Decision Support Systems

Business Intelligence and Analytics Systems for Decision Support TENTH EDITION

Ramesh Sharda • Dursun Delen • Efraim Turban

PEARSONA L W A Y S L E A R N I N G

T e n t h E d i t i o n

B u s i n e s s I n t e l l i g e n c e

a n d A n a l y t i c s :

S y s t e m s f o r D e c i s i o n S u p p o r t

Global Edition

Ramesh Sharda Oklahoma State University

Dursun Delen Oklahoma State University

Efraim Turban University o f Hawaii

With contributions by

J . E. A ronson The University o f Georgia

T ing-Peng Liang National Sun Yat-sen University

David King JDA Software Group, Inc.

PEARSON B o s t o n C o lu m b u s I n d ia n a p o l is N e w Y o r k S a n F r a n c i s c o U p p e r S a d d le R iv e r

A m s te r d a m C a p e T o w n D u b a i L o n d o n M a d rid M ila n M u n i c h P a r is M o n t r e a l T o r o n t o D e l h i M e x i c o C ity S a o P a u l o S y d n e y H o n g K o n g S e o u l S i n g a p o r e T a i p e i T o k y o

Editor in Chief: Stephanie Wall Executive Editor: Bob Horan P ublisher, Global Edition: Laura Dent Senior Acquisitions Editor, Global Edition: Steven

Jackson Program M anager Team Lead: Ashley Santora Program M anager: Denise Vaughn Marketing Manager, International: Kristin Schneider P ro ject M anager Team Lead: Judy Leale P ro ject M anager: Tom Benfatti Assistant P roject Editor, Global Edition: Paromita

Baneijee

Pearson Education Limited Edinburgh Gate Harlow Essex CM20 2JE England and Associated Companies throughout the world Visit us on the World Wide Web at: www.pearsonglobaleditions.com

© Pearson Education Limited 2014 The rights of Ramesh Sharda, Dursun Delen, and Efraim Turban to be identified as the authors of this work have been asserted by them in accordance with the Copyright, Designs, and Patents Act 1988. Authorized adaptation f r o m the United States edition, entitled Business Intelligence a n d Analytics: Systems f o r D ecision Support, l ( f h edition, ISBN978-0-133-05090-5, by Ramesh Sharda, Dursun Delen, a n d E fraim Turban, p u b lish ed by P earson E ducation © 2014. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmittedin any form or by any means, electronic, mechanical, photocopying, recording or otheiwise, withouteither the prior written permission o f the publisher or a license permitting restricted copying in the United Kingdom issued by the Copyright Licensing Agency Ltd, Saffron House, 6-10 Kirby Street, London EC1N 8TS. All trademarks used herein are the property o f their respective owners.The use of any trademark in this text does not vest in the author or publisher any trademark ownership rights in such trademarks, nor does the use of such trademarks imply any affiliation with or endorsement of this book by such owners. Microsoft and/or its respective suppliers make no representations about the suitability of the information contained in the documents and related graphics published as part of the services for any purpose. All such documents and related graphics are provided “as is” without warranty o f any kind. Microsoft and/or its respective suppliers hereby disclaim all warranties and conditions with regard to this information, including all warranties and conditions of merchantability, whether express, implied or statutory, fitness for a particular purpose, title and non-infringement. In no event shall Microsoft and/or its respective suppliers be liable for any special, indirect or consequential damages or any damages whatsoever resulting from loss of use, data or profits, whether in an action of contract, negligence or other tortious action, arising out o f or in connection with the use or performance o f information available from the sendees. The documents and related graphics contained herein could include technical inaccuracies or typographical errors. Changes are periodically added to the information herein. Microsoft and/or its respective suppliers may make improvements and/or changes in the product(s) and/or the program(s) described herein at any time. Partial screen shots may be viewed in hill within the software version specified. Microsoft® and Windows® are registered trademarks of the Microsoft Corporation in the U.S.A. and other countries. This book is not sponsored or endorsed by or affiliated with the Microsoft Corporation.

ISBN 10: 1-292-00920-9 ISBN 13: 978-1-292-00920-9 British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library

9 8 7 6 5 4 3 2 1 14 13 12 11 10 Typeset in ITC Garamond Std. Integra Software Solutions Printed and bound by Courier Kendalville in The United States o f America

Operations Specialist: Michelle Klein Senior Manufacturing Controller, Production: Trudy

Kimber Creative D irector: Jayne Conte Cover Im age Credit: © Robert Adrian Hillman Cover P rin ter: Courier Kendallville Cover D esig n er: Jodi Notowitz at Wicked Design Full-Service P ro je c t Managem ent: George Jacob,

Integra Software Solutions. T ex t Font: ITC Garamond Std

BRIEF CONTENTS

Preface 21 A b o u t t h e A u t h o r s 2 9

P A R T I D e c is io n M a k in g a n d A n a ly t ic s : A n O v e r v ie w 31 C h a p te r 1 A n O v e r v ie w o f Bu sin ess In te llig e n c e , A n a ly tic s ,

a n d D e cisio n S u p p o r t 32 C h a p te r 2 F o u n d a tio n s a n d T e c h n o lo g ie s f o r D e c is io n M a k in g 67

P A R T II D e s c r ip t iv e A n a l y t i c s 107 C h a p te r 3 D a ta W a r e h o u s in g 108 C h a p te r 4 Bu sin e ss R e p o rtin g , V is u a l A n a ly tic s , a n d Business

P e r fo r m a n c e M a n a g e m e n t 165

P A R T III P r e d ic t iv e A n a l y t i c s 2 1 5 C h a p te r 5 D a ta M in in g 216

C h a p te r 6 T e c h n iq u e s f o r P re d ic tiv e M o d e lin g 273 C h a p te r 7 T e x t A n a ly tic s , T e x t M in in g , a n d S e n t im e n t A n a ly s is 318 C h a p te r 8 W e b A n a ly tic s , W e b M in in g , a n d S o c ia l A n a ly tic s 368

P A R T IV P r e s c r ip t iv e A n a l y t i c s 421 C h a p te r 9 M o d e l- B a s e d D e c is io n M a k in g : O p tim iz a tio n a n d M u lti-

C rite ria S ystem s 422 C h a p te r 10 M o d e lin g a n d A n a lys is: H e u ris tic S e a rc h M e t h o d s a n d

S im u la t io n 465 C h a p te r 11 A u t o m a t e d D e cisio n System s a n d E x p e rt S ystem s 499 C h a p te r 12 K n o w le d g e M a n a g e m e n t a n d C o lla b o r a tiv e System s 537

P A R T V B ig D a ta a n d F u t u r e D ir e c t io n s f o r B u s in e s s A n a l y t i c s 571 C h a p te r 13 B ig D a ta a n d A n a ly tic s 572 C h a p te r 14 Bu sin ess A n a ly tic s : E m e rg in g T re n d s a n d F u tu re

Im p a c ts 622

Glossary 664 Index 678

CONTENTS

Preface 21 A b o u t the A u th o rs 29

P a r t I D e c is io n M a k in g a n d A n a l y t ic s : A n O v e r v ie w 31 C h a p te r 1 An O verview o f Business Intelligence, Analytics, and

Decision Support 32 1.1 O p e n in g V ig n e t t e : M a g p ie S e n s in g E m p lo y s A n a ly tic s

M a n a g e a V a c c in e S u p p ly C h a in E ff e c tiv e ly a n d S a fe ly

1.2 C h a n g in g Bu sin ess E n v ir o n m e n ts a n d C o m p u te r iz e d D e cis io n S u p p o r t 35 The Business Pressures-Responses-Support Model 35

1.3 M a n a g e r ia l D e c is io n M a k in g 37 The Nature of Managers' Work 37 The Decision-Making Process 38

1.4 In fo r m a tio n S ystem s S u p p o r t f o r D e cis io n M a k in g 3 1.5 A n E a r ly F r a m e w o r k f o r C o m p u te r iz e d D ecisio n

S u p p o r t 41 The Gorry and Scott-Morton Classical Framework 41 Computer Support for Structured Decisions 42

Computer Support for Unstructured Decisions 43 Computer Support for Semistructured Problems 43

1.6 T h e C o n c e p t o f D e c is io n S u p p o r t S ys te m s (D S S ) 43 DSS as an Umbrella Term 43 Evolution of DSS into Business Intelligence 44

1.7 A F r a m e w o r k f o r Bu sin e ss In te llig e n c e (B l) 44 Definitions of Bl 44 A Brief History of Bl 44 The Architecture of Bl 45 Styles of Bl 45 The Origins and Drivers of Bl 46 A Multimedia Exercise in Business Intelligence 46 ► APPLICATION CASE 1.1 Sabre Helps Its Clients Through

and Analytics 47 The DSS-BI Connection 48

1.8 Bu sin e ss A n a ly tic s O v e r v ie w 49 Descriptive Analytics 50 ► APPLICATION CASE 1.2 Eliminating Inefficiencies at Seattle

Children's Hospital 51 ► APPLICATION CASE 1.3 Analysis a t the Speed of Thought 53 Predictive Analytics 52

Contents 5

► APPLICATION CASE 1.4 Moneyball: Analytics in Sports and Movies 53 ► APPLICATION CASE 1.5 Analyzing Athletic Injuries 54 Prescriptive Analytics 54 ► APPLICATION CASE 1.6 Industrial and Commercial Bank of China

(ICBC) Employs Models to Reconfigure Its Branch Network 55 Analytics Applied to Different Domains 56 Analytics or Data Science? 56

1.9 B r ie f In tr o d u c tio n t o B ig D a ta A n a ly tic s 57 What Is Big Data? 57 ► APPLICATION CASE 1.7 Gilt Groupe's Flash Sales Streamlined by Big

Data Analytics 59 1.10 P la n o f t h e B o o k 59

Part I: Business Analytics: An Overview 59 Part II: Descriptive Analytics 60 Part III: Predictive Analytics 60 Part IV: Prescriptive Analytics 61 Part V: Big Data and Future Directions for Business Analytics 61

1.11 R eso u rce s, Links, a n d t h e T e r a d a ta U n iv e r s ity N e tw o r k C o n n e c tio n 61 Resources and Links 61 Vendors, Products, and Demos 61 Periodicals 61 The Teradata University Network Connection 62 The Book's Web Site 62 Ch apter H ig h lig hts 62 • K ey Terms 63 Q uestions fo r Discussion 63 • Exercises 63 ► END-OF-CHAPTER APPLICATION CASE Nationwide Insurance Used Bl

to Enhance Customer Service 64 References 65

C h a p te r 2 Foundations and Technologies for Decision M aking 67 2.1 O p e n in g V ig n e t t e : D e c is io n M o d e lin g a t H P U sin g

S p r e a d s h e e ts 68 2.2 D e cisio n M a k in g : In tr o d u c tio n a n d D e fin itio n s 70

Characteristics of Decision Making 70 A Working Definition of Decision Making 71 Decision-Making Disciplines 71 Decision Style and Decision Makers 71

2.3 P h a se s o f t h e D e c is io n - M a k in g Process 72 2.4 D e c is io n M a k in g : T h e In te llig e n c e P h a se 74

Problem (or Opportunity) Identification 75 ► APPLICATION CASE 2.1 Making Elevators Go Faster! 75

Problem Classification 76 Problem Decomposition 76 Problem Ownership 76

2.5 D e c is io n M a k in g : T h e D e sig n P h a s e 77

Models 77 Mathematical (Quantitative) Models 77 The Benefits of Models 77 Selection of a Principle of Choice 78 Normative Models 79 Suboptimization 79 Descriptive Models 80 Good Enough, or Satisficing 81 Developing (Generating) Alternatives 82 Measuring Outcomes 83

Risk 83 Scenarios 84 Possible Scenarios 84 Errors in Decision Making 84

2.6 D e c is io n M a k in g : T h e C h o ic e P h a s e 85 2.7 D e cisio n M a k in g : T h e Im p le m e n ta tio n P h a s e 85

2.8 H o w D e cisio n s A r e S u p p o r te d 86 Support for the Intelligence Phase 86 Support for the Design Phase 87 Support for the Choice Phase 88 Support for the Implementation Phase 88

2.9 D e c is io n S u p p o r t System s: C a p a b ilitie s 89 A DSS Application 89

2.10 DSS C la ss ific a tio n s 91 The AIS SIGDSS Classification for DSS 91

Other DSS Categories 93 Custom-Made. S'fste.ccvs, Vex'.us Read^-Wlade Systems 93

2.11 C o m p o n e n ts o f D e c is io n S u p p o r t S ys te m s 94 The Data Management Subsystem 95 The Model Management Subsystem 95 ► APPLICATION CASE 2.2 Station Casinos Wins by Building Customer

Relationships Using Its Data 96 ► APPLICATION CASE 2.3 SNAP DSS Helps OneNet Make

Telecommunications Rate Decisions 98

The User Interface Subsystem 98 The Knowledge-Based Management Subsystem 99 ► APPLICATION CASE 2.4 From a Game Winner to a Doctor! 100 C h ap ter H ig h lig hts 102 • K e y Terms 103 Questions fo r Discussion 103 • Exercises 104 ► END-OF-CHAPTER APPLICATION CASE Logistics Optimization in a

Major Shipping Company (CSAV) 104 References 105

P a r t II D e s c r ip t iv e A n a l y t i c s 1 07 C h a p te r 3 Data W arehousing 108

3.1 O p e n in g V ig n e t t e : Isle o f C a p ri C asin os Is W i n n in g w it h E n te rp ris e D a ta W a r e h o u s e 109

3.2 D a ta W a r e h o u s in g D e fin itio n s a n d C o n c e p ts 111 What Is a Data Warehouse? 111

A Historical Perspective to Data Warehousing 111 Characteristic of Data Warehousing 113 Data Marts 114 Operational Data Stores 114

Enterprise Data Warehouses (EDW) 115 Metadata 115 ► APPLICATION CASE 3.1 A Better Data Plan: Well-Established TELCOs

Leverage Data Warehousing and Analytics to Stay on Top in a Competitive Industry 115

3.3 D a ta W a r e h o u s in g Process O v e r v ie w 117 ► APPLICATION CASE 3.2 Data Warehousing Helps MultiCare Save

More Lives 118

3.4 D a ta W a r e h o u s in g A r c h ite c tu r e s 120 Alternative Data Warehousing Architectures 123 Which Architecture Is the Best? 126

3.5 D a ta In te g r a tio n a n d t h e E x tra c tio n , T r a n s fo r m a tio n , a n d L o a d (E T L ) Processes 127 Data Integration 128 ► APPLICATION CASE 3.3 BP Lubricants Achieves BIGS Success 128 Extraction, Transformation, and Load 130

3.6 D a ta W a r e h o u s e D e v e lo p m e n t 132 ► APPLICATION CASE 3.4 Things Go Better with Coke's Data

Warehouse 133

Data Warehouse Development Approaches 133 ► APPLICATION CASE 3.5 Starwood Hotels & Resorts Manages Hotel

Profitability with Data Warehousing 136

Additional Data Warehouse Development Considerations 137 Representation of Data in Data Warehouse 138 Analysis of Data in the Data Warehouse 139 OLAP Versus OLTP 140 0LAP Operations 140

3.7 D a ta W a r e h o u s in g Im p le m e n ta tio n Issues 143 ► APPLICATION CASE 3.6 EDW Helps Connect State Agencies in

Michigan 145

Massive Data Warehouses and Scalability 146

3.8 R eal-T im e D a ta W a r e h o u s in g 147 ► APPLICATION CASE 3.7 Egg Pic Fries the Competition in Near Real

Time 148

3.9 D a ta W a r e h o u s e A d m in is tr a tio n , S e c u r ity Issues, a n d Fut_-~ T re n d s 151 The Future of Data Warehousing 153

3.10 R eso u rce s, Links, a n d t h e T e r a d a ta U n iv e rs ity N e tw o r k C o n n e c tio n 156

Resources and Links 156 Cases 156 Vendors, Products, and Demos 157 Periodicals 157 Additional References 157 The Teradata University Network (TUN) Connection 157 Ch apter H ig h lig hts 158 • K e y Terms 158 Questions f o r Discussion 158 • Exercises 159 ► END-OF-CHAPTER APPLICATION CASE Continental Airlines Flies High

with Its Real-Time Data Warehouse 161 R eferences 162

C h a p te r 4 Business Reporting, Visual Analytics, and Business Perform ance M anagem ent 165

4.1 O p e n in g V ig n e tte :S e lf- S e rv ic e R e p o rtin g E n v ir o n m e n t S a ve s M illio n s f o r C o r p o r a te C u sto m e rs 166

4.2 Bu sin ess R e p o rtin g D e fin itio n s a n d C o n c e p ts 169 What Is a Business Report? 170 ► APPLICATION CASE 4.1 Delta Lloyd Group Ensures Accuracy and

Efficiency in Financial Reporting 171 Components of the Business Reporting System 173 ► APPLICATION CASE 4.2 Flood of Paper Ends at FEMA 174

4.3 D a ta a n d In fo r m a tio n V is u a liz a tio n 175 ► APPLICATION CASE 4.3 Tableau Saves Blastrac Thousands of Dollars

with Simplified Information Sharing 176 A Brief History of Data Visualization 177 ► APPLICATION CASE 4.4 TIBCO Spotfire Provides Dana-Farber Cancer

Institute with Unprecedented Insight into Cancer Vaccine Clinical Trials 179

4.4 D if f e r e n t T yp e s o f C h a rts a n d G r a p h s 180 Basic Charts and Graphs 180

Spedalized Charts and Graphs 181 4.5 T h e E m e rg e n c e o f D a ta V is u a liz a tio n a n d V is u a l

A n a ly tic s 184 Visual Analytics 186 High-Powered Visual Analytics Environments 188

4.6 P e r fo r m a n c e D a s h b o a rd s 190 ► APPLICATION CASE 4.5 Dallas Cowboys Score Big with Tableau and

Teknion 191

Dashboard Design 192 ► APPLICATION CASE 4.6 Saudi Telecom Company Excels with

Information Visualization 193

What to Look For in a Dashboard 194

Best Practices in Dashboard Design 195 Benchmark Key Performance Indicators with Industry Standards 195

Wrap the Dashboard Metrics with Contextual Metadata 195

Validate the Dashboard Design by a Usability Specialist 195

Prioritize and Rank Alerts/Exceptions Streamed to the Dashboard 195

Enrich Dashboard with Business Users' Comments 195

Present Information in Three Different Levels 196 Pick the Right Visual Construct Using Dashboard Design Principles 196

Provide for Guided Analytics 196 4.7 Bu sin ess P e r fo r m a n c e M a n a g e m e n t 196

Closed-Loop BPM Cycle 197 ► APPLICATION CASE 4.7 IBM Cognos Express Helps Mace for Faster

and Better Business Reporting 199

4.8 P e r fo r m a n c e M e a s u r e m e n t 200 Key Performance Indicator (KPI) 201

Performance Measurement System 202 4.9 B a la n c e d S c o re c a rd s 202

The Four Perspectives 203

The Meaning of Balance in BSC 204

Dashboards Versus Scorecards 204 4.10 Six S ig m a as a P e r fo r m a n c e M e a s u r e m e n t S y s te m 205

The DMAIC Performance Model 206

Balanced Scorecard Versus Six Sigma 206

Effective Performance Measurement 207 ► APPLICATION CASE 4.8 E xpedia.com 's Customer Satisfaction

Scorecard 208 Ch apter H ig h lig hts 209 • K ey Terms 210 Q uestions fo r Discussion 211 • Exercises 211 ► END-OF-CHAPTER APPLICATION CASE Smart Business Reporting

Helps Healthcare Providers Deliver Better Care 212 R eferences 214

P a r t III P r e d ic t iv e A n a l y t i c s 2 1 5 C h a p te r 5 Data M ining 216

5.1 O p e n in g V ig n e t t e : C a b e la 's R e e ls in M o r e C u s to m e rs w it h A d v a n c e d A n a ly tic s a n d D a ta M in in g 217

5.2 D a ta M in in g C o n c e p ts a n d A p p lic a tio n s 219 ► APPLICATION CASE 5.1 Smarter Insurance: Infinity P&C Improves

Customer Service and Combats Fraud with Predictive Analytics 221

1 0 Contents

Definitions, Characteristics, and Benefits 222 ► APPLICATION CASE 5.2 Harnessing Analytics to Combat Crime:

Predictive Analytics Helps Memphis Police Department Pinpoint Crime and Focus Police Resources 226

How Data Mining Works 227 Data Mining Versus Statistics 230

5.3 D a ta M in in g A p p lic a tio n s 231 ► APPLICATION CASE 5.3 A Mine on Terrorist Funding 233

5.4 D a ta M in in g Pro cess 234 Step 1: Business Understanding 235 Step 2: Data Understanding 235 Step 3: Data Preparation 236 Step 4: Model Building 238 ► APPLICATION CASE 5.4 Data Mining in Cancer Research 240 Step 5: Testing and Evaluation 241 Step 6: Deployment 241 Other Data Mining Standardized Processes and Methodologies 242

5.5 D a ta M in in g M e th o d s 244 Classification 244 Estimating the True Accuracy of Classification Models 245 Cluster Analysis for Data Mining 250 ► APPLICATION CASE 5.5 2degrees Gets a 1275 Percent Boost in Churn

Identification 251 Association Rule Mining 254

5.6 D a ta M in in g S o f t w a r e T o o ls 258 ► APPLICATION CASE 5.6 Data Mining Goes to Hollywood: Predicting

Financial Success of Movies 261 5.7 D a ta M in in g P riv a c y Issues, M y th s , a n d B lu n d e rs 264

Data Mining and Privacy Issues 264 ► APPLICATION CASE 5.7 Predicting Customer Buying Patterns— The

Target Story 265 Data Mining Myths and Blunders 266 C h ap ter H ig h lig h ts 267 • K e y Terms 268 Questions f o r Discussion 2 6 8 • Exercises 269 ► END-OF-CHAPTER APPLICATION CASE Macys.com Enhances Its

Customers' Shopping Experience with Analytics 271 R eferences 271

C h a p te r 6 Techniques fo r Predictive M odeling 273 6.1 O p e n in g V ig n e t t e : P r e d ic t iv e M o d e lin g H e lp s B e t t e r

U n d e rs ta n d a n d M a n a g e C o m p le x M e d ic a l P ro c e d u re s 274

6.2 Ba sic C o n c e p ts o f N e u r a l N e tw o r k s 277 Biological and Artificial Neural Networks 278 ► APPLICATION CASE 6.1 Neural Networks Are Helping to Save Lives in

the Mining Industry 280 Elements of ANN 281

C ontents 11

Network Information Processing 282 Neural Network Architectures 284 ► APPLICATION CASE 6.2 Predictive Modeling Is Powering the Power

Generators 286 6.3 D e v e lo p in g N e u ra l N e tw o r k - B a s e d S ystem s 288

The General ANN Learning Process 289 Backpropagation 290

6.4 Illu m in a tin g t h e B la c k B o x o f A N N w it h S e n s itiv ity A n a ly s is 292 ► APPLICATION CASE 6.3 Sensitivity Analysis Reveals Injury Severity

Factors in Traffic Accidents 294

6.5 S u p p o r t V e c t o r M a c h in e s 295 ► APPLICATION CASE 6.4 Managing Student Retention with Predictive

Modeling 296 Mathematical Formulation of SVMs 300 Primal Form 301 Dual Form 301 Soft Margin 301 Nonlinear Classification 302 Kernel Trick 302

6.6 A Process-Based A p p r o a c h t o t h e U se o f S V M 303 Support Vector Machines Versus Artificial Neural Networks 304

6.7 N e a r e s t N e ig h b o r M e t h o d f o r P re d ic tio n 305 Similarity Measure: The Distance Metric 306 Parameter Selection 307 ► APPLICATION CASE 6.5 Efficient Image Recognition and

Categorization with kNN 308 Ch ap ter H ighlights 310 • K ey Terms 310 Q uestions fo r Discussion 311 • Exercises 311 ► END-OF-CHAPTER APPLICATION CASE Coors Improves Beer Flavors

with Neural Networks 314 R eferences 315

C h a p te r 7 Text Analytics, Text M ining, and Sentim ent Analysis 318 7.1 O p e n in g V ig n e t t e : M a c h in e V e rsu s M e n o n Jeo pa rdy7: T h e

S to ry o f W a t s o n 319 7.2 T e x t A n a ly tic s a n d T e x t M in in g C o n c e p ts a n d

D e fin itio n s 321 ► APPLICATION CASE 7.1 Text Mining for Patent Analysis 325

7.3 N a tu r a l L a n g u a g e P ro ce ssin g 326 ► APPLICATION CASE 7.2 Text Mining Improves Hong Kong

Government's Ability to Anticipate and Address Public Complaints 328

7.4 T e x t M in in g A p p lic a tio n s 330 Marketing Applications 331 Security Applications 331 ► APPLICATION CASE 7.3 Mining for Lies 332

Biomedical Applications 334

1 2 Contents

Academic Applications 335 ► APPLICATION CASE 7.4 Text Mining and Sentiment Analysis Help

Improve Customer Service Performance 336 7.5 T e x t M in in g Pro cess 337

Task 1: Establish the Corpus 338

Task 2; Create the Term-Document Matrix 339

. Task 3: Extract the Knowledge 342 ► APPLICATION CASE 7.5 Research Literature Survey with Text

Mining 344 7.6 T e x t M in in g T o o ls 347

Commercial Software Tools 347

Free Software Tools 347 ► APPLICATION CASE 7.6 A Potpourri of Text Mining Case Synopses 348

7.7 S e n t im e n t A n a ly s is O v e r v ie w 349 ► APPLICATION CASE 7.7 Whirlpool Achieves Customer Loyalty and

Product Success with Text Analytics 351 7.8 S e n t im e n t A n a ly s is A p p lic a tio n s 353 7.9 S e n t im e n t A n a ly s is Process 355

Methods for Polarity Identification 356

Using a Lexicon 357

Using a Collection of Training Documents 358 Identifying Semantic Orientation of Sentences and Phrases 358

Identifying Semantic Orientation of Document 358

7.10 S e n t im e n t A n a ly s is a n d S p e e c h A n a ly tic s 359 How Is It Done? 359 ► APPLICATION CASE 7.8 Cutting Through the Confusion: Blue Cross

Blue Shield of North Carolina Uses Nexidia's Speech Analytics to Ease Member Experience in Healthcare 361

C h apter H ig h lig hts 363 • K e y Terms 363 Q uestions fo r Discussion 364 • Exercises 364 ► END-OF-CHAPTER APPLICATION CASE BBVA Seamlessly Monitors

and Improves Its Online Reputation 365 References 366

C h a p te r 8 W eb Analytics, W eb M ining, and Social Analytics 368 8.1 O p e n in g V ig n e t t e : S e c u r ity F irs t In s u ra n c e D e e p e n s

C o n n e c tio n w it h P o lic y h o ld e rs 369 8.2 W e b M in in g O v e r v ie w 371 8.3 W e b C o n te n t a n d W e b S tr u c tu r e M in in g 374

► APPLICATION CASE 8.1 Identifying Extremist Groups with Web Link and Content Analysis 376

8.4 S e a rc h E n g in e s 377 Anatomy of a Search Engine 377 1. Development Cycle 378 Web Crawler 378 Document Indexer 378

2. Response Cycle 379 Query Analyzer 379 Document Matcher/Ranker 379 How Does Google Do It? 381 ► APPLICATION CASE 8.2 IGN Increases Search Traffic by 1500 Percent 383

8.5 S e a rc h E n g in e O p tim iz a tio n 384 Methods for Search Engine Optimization 385 ► APPLICATION CASE 8.3 Understanding Why Customers Abandon

Shopping Carts Results in $10 Million Sales Increase 387

8.6 W e b U s a g e M in in g ( W e b A n a ly tic s ) 388 Web Analytics Technologies 389 ► APPLICATION CASE 8.4 Allegro Boosts Online Click-Through Rates by

500 Percent with Web Analysis 390 Web Analytics Metrics 392 Web Site Usability 392 Traffic Sources 393 Visitor Profiles 394 Conversion Statistics 394

8.7 W e b A n a ly tic s M a t u r it y M o d e l a n d W e b A n a ly tic s T o o ls 396 Web Analytics Tools 398 Putting It All Together— A Web Site Optimization Ecosystem 400 A Framework for Voice of the Customer Strategy 402

8.8 S o c ia l A n a ly tic s a n d S o c ia l N e t w o r k A n a ly s is 403 Social Network Analysis 404 Social Network Analysis Metrics 405 ► APPLICATION CASE 8.5 Social Network Analysis Helps

Telecommunication Firms 405 Connections 406 Distributions 406 Segmentation 407

8.9 S o c ia l M e d ia D e fin itio n s a n d C o n c e p ts 407 How Do People Use Social Media? 408 ► APPLICATION CASE 8.6 Measuring the Impact of Social Media at

Lollapalooza 409

8.10 S o c ia l M e d ia A n a ly tic s 410 Measuring the Social Media Impact 411 Best Practices in Social Media Analytics 411 ► APPLICATION CASE 8.7 eHarmony Uses Social Media to Help Take the

Mystery Out of Online Dating 413 Social Media Analytics Tools and Vendors 414 Ch apter H ig h lig hts 416 • K e y Terms 417 Q uestions fo r Discussion 417 • Exercises 418 ► END-OF-CHAPTER APPLICATION CASE Keeping Students on Track with

Web and Predictive Analytics 418 References 420

P a r t IV P r e s c r ip t iv e A n a l y t i c s 421 C h a p te r 9 Model-Based Decision M aking: O ptim ization and

M ulti-Criteria System s 422 9.1 O p e n in g V ig n e t t e : M id w e s t IS O S a ve s B illio n s b y B e t t e r

P la n n in g o f P o w e r P la n t O p e r a tio n s a n d C a p a c ity P la n n in g 423

9.2 D e c is io n S u p p o r t System s M o d e lin g 424 ► APPLICATION CASE 9.1 Optimal Transport for ExxonMobil

Downstream Through a DSS 425 Current Modeling Issues 426 ► APPLICATION CASE 9.2 Forecasting/Predictive Analytics Proves to Be

a Good Gamble for Harrah's Cherokee Casino and Hotel 427

9.3 S tru c tu re o f M a th e m a tic a l M o d e ls f o r D ecision S u p p o rt 429 The Components of Decision Support Mathematical Models 429 The Structure of Mathematical Models 431

9.4 C e r ta in ty , U n c e r ta in ty , a n d R isk 431 Decision Making Linder Certainty 432 Decision Making Under Uncertainty 432 Decision Making Under Risk (Risk Analysis) 432 ► APPLICATION CASE 9.3 American Airlines Uses

Should-Cost Modeling to Assess the Uncertainty of Bids for Shipment Routes 433

9.5 D e c is io n M o d e lin g w it h S p r e a d s h e e ts 434 ► APPLICATION CASE 9.4 Showcase Scheduling at Fred Astaire East

Side Dance Studio 434

9.6 M a t h e m a t ic a l P ro g ra m m in g O p tim iz a tio n 437 ► APPLICATION CASE 9.5 Spreadsheet Model Helps Assign Medical

Residents 437

Mathematical Programming 438 Linear Programming 438 Modeling in LP: An Example 439 Implementation 444

9.7 M u lt ip le G o a ls, S e n s itiv ity A n a ly s is , W h a t - lf A n alysis, a n d G o a l S e e k in g 446 Multiple Goals 446 Sensitivity Analysis 447 What-lf Analysis 448 Goal Seeking 448

9.8 D e cisio n A n a ly s is w it h D e c is io n T a b le s a n d D e cisio n T re e s 450 Decision Tables 450 Decision Trees 452

9.9 M u lti- C rite ria D e cis io n M a k in g W i t h P a irw is e C o m p a ris o n s 453 The Analytic Hierarchy Process 453

► APPLICATION CASE 9.6 U.S. HUD Saves the House by Using AHP for Selecting IT Projects 453

Tutorial on Applying Analytic Hierarchy Process Using Web-HIPRE 455 Ch apter H ig h lig hts 459 • K e y Terms 460 Q uestions fo r Discussion 460 • Exercises 460 ► END-OF-CHAPTER APPLICATION CASE Pre-Positioning of Emergency

Items for CARE International 463 References 464

Chapter 10 M odeling and Analysis: Heuristic Search M ethods and Sim ulation 465 10.1 O p e n in g V ig n e t t e : S ys te m D y n a m ic s A llo w s F lu o r

C o rp o ra tio n t o B e t t e r P la n f o r P ro je c t a n d C h a n g e M a n a g e m e n t 466

10.2 P ro b le m - S o lv in g S e a rc h M e th o d s 467 Analytical Techniques 468 Algorithms 468 Blind Searching 469

Heuristic Searching 469 ► APPLICATION CASE 10.1 Chilean Government Uses Heuristics to

Make Decisions on School Lunch Providers 469 10.3 G e n e tic A lg o r ith m s a n d D e v e lo p in g G A A p p lic a tio n s 471

Example: The Vector Game 471 Terminology of Genetic Algorithms 473 How Do Genetic Algorithms Work? 473 Limitations of Genetic Algorithms 475

Genetic Algorithm Applications 475 10.4 S im u la t io n 476

► APPLICATION CASE 10.2 Improving Maintenance Decision Making in the Finnish Air Force Through Simulation 476

► APPLICATION CASE 10.3 Simulating Effects of Hepatitis B Interventions 477

Major Characteristics of Simulation 478 Advantages of Simulation 479 Disadvantages of Simulation 480 The Methodology of Simulation 480 Simulation Types 481 Monte Carlo Simulation 482 Discrete Event Simulation 483

10.5 V is u a l In te ra c tiv e S im u la t io n 483 Conventional Simulation Inadequacies 483 Visual Interactive Simulation 483 Visual Interactive Models and DSS 484 ► APPLICATION CASE 10.4 Improving Job-Shop Scheduling Decisions

Through RFID: A Simulation-Based Assessment 484 Simulation Software 487

10.6 S ys te m D y n a m ic s M o d e lin g 488 10.7 A g e n t- B a s e d M o d e lin g 491

► APPLICATION CASE 10.5 Agent-Based Simulation Helps Analyze Spread of a Pandemic Outbreak 493

C h apter H ighlights 494 • K e y Terms 494 Q uestions fo r Discussion 495 • Exercises 495 ► END-OF-CHAPTER APPLICATION CASE HP Applies Management

Science Modeling to Optimize Its Supply Chain and Wins a Major Award 495

R eferences 497

C h a p te r 11 Autom ated Decision System s and Expert System s 499 11.1 O p e n in g V ig n e t t e : In te r C o n tin e n ta l H o te l G r o u p Uses

D e c is io n R u le s f o r O p tim a l H o te l R o o m R a te s 500 11.2 A u t o m a t e d D e c is io n S y s te m s 501

► APPLICATION CASE 11.1 Giant Food Stores Prices the Entire Store 502

11.3 T h e A r tific ia l In te llig e n c e F ie ld 505 11.4 Ba sic C o n c e p ts o f E x p e rt S y s te m s 507

Experts 507 Expertise 508 Features of ES 508 ► APPLICATION CASE 11.2 Expert System Helps in Identifying Sport

Talents 510 11.5 Applications of Expert Systems 510

► APPLICATION CASE 11.3 Expert System Aids in Identification of Chemical, Biological, and Radiological Agents 511

Classical Applications of ES 511 Newer Applications of ES 512 Areas for ES Applications 513

11.6 S tru c tu r e o f E x p e rt System s 514 Knowledge Acquisition Subsystem 514 Knowledge Base 515 Inference Engine 515 User Interface 515 Blackboard (Workplace) 515 Explanation Subsystem (Justifier) 516 Knowledge-Refining System 516 ► APPLICATION CASE 11.4 Diagnosing Heart Diseases by Signal

Processing 516

11.7 K n o w le d g e E n g in e e r in g 517 Knowledge Acquisition 518 Knowledge Verification and Validation 520 Knowledge Representation 520 Inferencing 521 Explanation and Justification 526

Contents 17

11.8 P ro b le m A r e a s S u ita b le f o r E x p e rt S ys te m s 527 11.9 D e v e lo p m e n t o f E x p e r t S ystem s 528

Defining the Nature and Scope of the Problem 529 Identifying Proper Experts 529 Acquiring Knowledge 529 Selecting the Building Tools 529 Coding the System 531 Evaluating the System 531 ► APPLICATION CASE 11.5 Clinical Decision Support System for Tendon Injuries 531

11.10 C o n c lu d in g R e m a rk s 532 Ch apter H ig h lig hts 533 • K e y Terms 533 Q uestions fo r Discussion 534 • Exercises 534 ► END-OF-CHAPTER APPLICATION CASE Tax Collections Optimization

for New York State 534 References 535

C h a p te r 12 Know ledge M anagem ent and Collaborative System s 537 12.1 O p e n in g V ig n e t t e : E x p e rtis e T ra n s fe r S ys te m t o T ra in

F u tu r e A r m y P e r s o n n e l 538 12.2 Introduction to Knowledge Management 542

Knowledge Management Concepts and Definitions 543

Knowledge 543 Explicit and Tacit Knowledge 545

12.3 A p p r o a c h e s t o K n o w le d g e M a n a g e m e n t 546 The Process Approach to Knowledge Management 547 The Practice Approach to Knowledge Management 547 Hybrid Approaches to Knowledge Management 548 Knowledge Repositories 548

12.4 In fo r m a tio n T e c h n o lo g y (IT ) in K n o w le d g e M a n a g e m e n t 550 The KMS Cycle 550 Components of KMS 551 Technologies That Support Knowledge Management 551

12.5 M a k in g D e cisio n s in G ro u p s : C h a ra c te ris tic s, Process, B e n e fits , a n d D y s fu n c tio n s 553 Characteristics of Groupwork 553 The Group Decision-Making Process 554 The Benefits and Limitations of Groupwork 554

12.6 S u p p o r tin g G r o u p w o r k w it h C o m p u te r iz e d S ys te m s 556 An Overview of Group Support Systems (GSS) 556 Groupware 557 Time/Place Framework 557

12.7 T o o ls f o r In d ire c t S u p p o r t o f D e c is io n M a k in g 558 Groupware Tools 558

Contents

Groupware 560 Collaborative Workflow 560 Web 2.0 560 Wikis 561 Collaborative Networks 561

12.8 D ire c t C o m p u te r iz e d S u p p o r t f o r D e c is io n M a k in g : Fro m G r o u p D e cis io n S u p p o r t S ys te m s t o G r o u p S u p p o rt S ys te m s 562 Group Decision Support Systems (GDSS) 562 Group Support Systems 563 How GDSS (or GSS) Improve Groupwork 563 Facilities for GDSS 564 Ch apter H ig h lig hts 565 • K ey Terms 566 Questions f o r Discussion 566 • Exercises 566 ► END-OF-CHAPTER APPLICATION CASE Solving Crimes by Sharing

Digital Forensic Knowledge 567 References 569

P a r t V B ig D a ta a n d F u t u r e D ir e c t io n s f o r B u s in e s s A n a l y t i c s 571

C h a p te r 13 Big Data and A nalytics 572 13.1 O p e n in g V ig n e t t e : B ig D a ta M e e ts B ig S c ie n c e a t C E R N 573 13.2 D e fin itio n o f B ig D a ta 576

The Vs That Define Big Data 577 ► APPLICATION CASE 13.1 Big Data Analytics Helps Luxottica Improve

Its Marketing Effectiveness 580 13.3 F u n d a m e n ta ls o f B ig D a ta A n a ly tic s 581

Business Problems Addressed by Big Data Analytics 584 ► APPLICATION CASE 13.2 Top 5 Investment Bank Achieves Single

Source of Truth 585 13.4 B ig D a ta T e c h n o lo g ie s 586

MapReduce 587 Why Use MapReduce? 588 Hadoop 588 How Does Hadoop Work? 588 Hadoop Technical Components 589 Hadoop: The Pros and Cons 590 NoSQL 592 ► APPLICATION CASE 13.3 eBay's Big Data Solution 593

13.5 D a ta S c ie n tis t 595 Where Do Data Scientists Come From? 595 ► APPLICATION CASE 13.4 Big Data and Analytics in Politics 598

13.6 B ig D a ta a n d D a ta W a r e h o u s in g 599 Use Case(s) for Hadoop 600 Use Case(s) for Data Warehousing 601

The Gray Areas (Any One of the Two Would Do the Job) 602 Coexistence of Hadoop and Data Warehouse 602

13.7 B ig D a ta V e n d o r s 604 ► APPLICATION CASE 13.5 Dublin City Council Is Leveraging Big Data

to Reduce Traffic Congestion 605 ► APPLICATION CASE 13.6 Creditreform Boosts Credit Rating Quality

with Big Data Visual Analytics 610 13.8 B ig D a ta a n d S tre a m A n a ly tic s 611

Stream Analytics Versus Perpetual Analytics 612 Critical Event Processing 612 Data Stream Mining 613

13.9 A p p lic a tio n s o f S tre a m A n a ly tic s 614 e-Commerce 614 Telecommunications 614 ► APPLICATION CASE 13.7 Turning Machine-Generated Streaming Data

into Valuable Business Insights 615 Law Enforcement and Cyber Security 616 Power Industry 617 Financial Services 617 Health Sciences 617 Government 617 Ch apter H ig h lig hts 618 • K e y Terms 618 Q uestions fo r Discussion 618 • Exercises 619 ► END-OF-CHAPTER APPLICATION CASE Discovery Health Turns Big

Data into Better Healthcare 619 R eferences 621

C h a p te r 14 Business Analytics: Em erging Trends and Future Impacts 622 14.1 O p e n in g V ig n e t t e : O k la h o m a G a s a n d E le c tr ic E m p lo y s

A n a ly tic s t o P r o m o te S m a r t E n e r g y U se 623 14.2 Lo c a tio n - B a s e d A n a ly tic s f o r O rg a n iz a tio n s 624

Geospatial Analytics 624 ► APPLICATION CASE 14.1 Great Clips Employs Spatial Analytics to

Shave Time in Location Decisions 626 A Multimedia Exercise in Analytics Employing Geospatial Analytics 627 Real-Time Location Intelligence 628 ► APPLICATION CASE 14.2 Quiznos Targets Customers for Its

Sandwiches 629 14.3 A n a ly tic s A p p lic a tio n s f o r C o n su m e rs 630

► APPLICATION CASE 14.3 A Life Coach in Your Pocket 631 14.4 R e c o m m e n d a tio n E n g in e s 633 14.5 W e b 2.0 a n d O n lin e S o c ia l N e t w o r k in g 634

Representative Characteristics of Web 2.0 635 Social Networking 635 A Definition and Basic Information 636 Implications of Business and Enterprise Social Networks 636

14.6 C lo u d C o m p u tin g a n d B l 637 Service-Oriented DSS 638 Data-as-a-Service (DaaS) 638 Information-as-a-Service (Information on Demand) (laaS) 641 Analytics-as-a-Service (AaaS) 641

14.7 Im p a c ts o f A n a ly tic s in O rg a n iz a tio n s : A n O v e r v ie w 643 New Organizational Units 643 Restructuring Business Processes and Virtual Teams 644 The Impacts of ADS Systems 644 Job Satisfaction 644 Job Stress and Anxiety 644 Analytics' Impact on Managers' Activities and Their Performance 645

14.8 Issues o f L e g a lity , P riva c y, a n d Eth ics 646 Legal Issues 646 Privacy 647 Recent Technology Issues in Privacy and Analytics 648 Ethics in Decision Making and Support 649

14.9 A n O v e r v ie w o f t h e A n a ly tic s Eco s ys te m 650 Analytics Industry Clusters 650 Data Infrastructure Providers 650 Data Warehouse Industry 651 Middleware Industry 652 Data Aggregators/Distributors 652 Analytics-Focused Software Developers 652 Reporting/Analytics 652 Predictive Analytics 653 Prescriptive Analytics 653 Application Developers or System Integrators: Industry Specific or General 654 Analytics User Organizations 655 Analytics Industry Analysts and Influences 657 Academic Providers and Certification Agencies 658 C h ap ter H ig h lig hts 659 • K e y Terms 659 Questions f o r Discussion 659 • Exercises 660 ► END-OF-CHAPTER APPLICATION CASE Southern States Cooperative

Optimizes Its Catalog Campaign 660 References 662

Glossary 664 In d e x 678

Overview of Business Intelligence, Analytics, and Decision Support

LEARNING OBJECTIVES

■ U n d erstand to d ay ’s tu rb u len t b u sin ess e n v iro n m en t an d d e s c rib e h o w org an izatio n s survive an d e v e n e x c e l in su ch a n e n v iro n m en t (so lv in g p ro b lem s a n d e x p lo itin g op p ortu n ities)

H U n d erstand th e n e e d fo r com p u terized su p p o rt o f m anagerial d e c is io n m aking

* U n d erstand a n early fram ew o rk for m an ag erial d e c is io n m aking

Th e b u sin e ss en v iro n m en t (clim a te ) is co n stan tly ch a n g in g , a n d it is b e c o m in g m o re and m o re co m p le x . O rgan izatio n s, private an d p u b lic, a re u n d e r p re ssu re s that fo rce th e m to re sp o n d q u ick ly to c h a n g in g co n d itio n s an d to b e in n ov ativ e in the w a y th ey o p era te. S u c h activities re q u ire o rg an izatio n s to b e a g ile an d to m a k e fre q u e n t a n d q u ick strateg ic, tactical, a n d o p era tio n a l d e cisio n s, s o m e o f w h ich are very co m p le x . M akin g s u ch d e cis io n s m ay re q u ire co n s id e ra b le am o u n ts o f relev an t data, in form ation, a n d k n o w le d g e. P ro ce s sin g th e s e , in th e fram ew o rk o f th e n e e d e d d e cisio n s, m u st b e d o n e q u ick ly , freq u en tly in real tim e, a n d u su ally req u ires s o m e co m p u te riz e d supp ort.

T h is b o o k is a b o u t u sin g b u s in e s s a n a ly tics as c o m p u te riz e d s u p p o rt fo r m a n a g e ­ rial d e c is io n m a k in g . It c o n c e n tr a te s o n b o t h th e th e o r e tic a l a n d c o n c e p tu a l fo u n d a ­ tio n s o f d e c is io n su p p o rt, as w e ll a s o n th e c o m m e rc ia l to o ls and te c h n iq u e s th a t are av a ila b le . T h is in tro d u cto ry c h a p te r p ro v id e s m o re d etails o f th e s e to p ic s a s w e ll a s a n o v e rv ie w o f th e b o o k . T h is c h a p te r h a s th e fo llo w in g s e c tio n s :

1 .1 O p e n in g V ig n e tte : M a g p ie S e n s in g E m p lo y s A n a ly tic s to M a n a g e a V a c c in e S u p p ly C h a in E ffe c tiv e ly a n d S a fe ly 3 3

1 .2 C h a n g in g B u s in e s s E n v ir o n m e n ts a n d C o m p u te riz e d D e c i s io n S u p p o rt 35

■ L earn th e c o n c e p tu a l fo u n d atio n s o f th e d e c is io n su p p o rt sy stem s (D S S 1) m e th o d o lo g y

■ D e s c r ib e th e b u sin e ss in te llig e n ce (B I) m e th o d o lo g y a n d c o n c e p ts and relate th e m to D SS

■ U nderstand th e various typ es o f analytics

■ List th e m a jo r to o ls o f co m p u terized d e c is io n su p p ort

‘The acronym DSS is treated as both singular and plural throughout this book. Similarly, other acronyms, such as MIS and GSS, designate both plural and singular forms. This is also true of the word analytics.

Chapter 1 • An O verview o f B u sin ess Intelligence, Analytics, a n d D ecision Support 3 3

1 .3 M a n a g e ria l D e c is io n M a k in g 3 7 1 .4 In fo r m a tio n S y s te m s S u p p o r t fo r D e c is io n M a k in g 3 9 1 .5 A n E a rly F ra m e w o rk fo r C o m p u te riz e d D e c is io n S u p p o rt 41 1 .6 T h e C o n c e p t o f D e c i s io n S u p p o rt S y s te m s (D S S ) 4 3 1 .7 A F ra m e w o rk fo r B u s in e s s I n te llig e n c e ( B I ) 4 4 1 .8 B u s in e s s A n a ly tic s O v e r v ie w 4 9 1 .9 B r i e f I n tr o d u c tio n to B ig D a ta A n a ly tics 5 7

1 .1 0 P la n o f th e B o o k 5 9 1 .1 1 R e s o u r c e s , L in k s, a n d th e T e r a d a ta U n iv ersity N e tw o rk C o n n e c tio n 61

1.1 OPENING VIGNETTE: Magpie Sensing Employs Analytics to Manage a Vaccine Supply Chain Effectively and Safely

C old ch a in in h e a lth ca re is d efin e d as th e te m p e ra tu re -co n tro lle d su p p ly ch a in involving a system o f tran sp ortin g and s to rin g v a c c in e s an d p h arm aceu tical drugs. It co n sists o f th ree m a jo r co m p o n e n ts — tran sp ort and sto rag e e q u ip m en t, trained p e rs o n n e l, an d e fficie n t m a n a g e m e n t p ro ced u res. T h e m ajority o f th e v a c c in e s in th e c o ld c h a in a re typ ically m a in ­ tain ed a t a tem p e ratu re o f 3 5 - 4 6 d eg ree s F ah re n h e it [ 2 - 8 d e g re e s C entigrade]. M aintaining cold ch a in integrity is e x trem ely im portant fo r h e a lth ca re p ro d u ct m anu factu rers.

E sp ecially fo r th e v a c c in e s , im p ro p er sto rag e a n d h an d lin g p ra ctice s th at co m p ro m ise v a ccin e viability p ro v e a co stly , tim e -co n su m in g affair. V a ccin e s m u st b e sto red p ro p e rly fro m m an u factu re until th e y a re a v ailab le fo r u se. Any e x tre m e te m p e ratu re s o f h e a t o r c o ld will re d u ce v a c c in e p o te n c y ; s u ch v a c c in e s , i f ad m inistered, m ight n o t y ield e ffe ctiv e results o r co u ld ca u se ad v e rse effects.

E ffectiv ely m aintainin g th e tem p eratu res o f sto rag e units th ro u g h o u t th e h e a lth ca re supp ly ch a in in re a l tim e— i.e ., b e g in n in g fro m th e gath erin g o f th e re so u rces, m a n u fa c­ turing, d istrib ution, a n d d isp e n sin g o f th e p rodu cts— is th e m o st e ffe ctiv e so lu tio n d esire d in th e c o ld ch ain . A lso, th e lo ca tio n -ta g g e d re al-tim e en v iro n m en tal data a b o u t the sto rag e units h e lp s in m o n ito rin g th e c o ld ch a in fo r s p o ile d p ro d u cts. T h e ch a in o f cu sto d y ca n b e easily id en tified to a ssig n p ro d u ct liability.

A stud y c o n d u cte d b y th e C enters fo r D ise a se C ontrol a n d P rev e n tio n (C D C ) lo o k e d a t the h an d lin g o f c o ld ch a in v a c c in e s b y 4 5 h e a lth ca re p ro vid ers a ro u n d U n ited States an d reported that th ree -q u arters o f th e p ro vid ers e x p e rie n c e d s erio u s c o ld c h a in v iolation s.

A WAY TOWARD A PO SSIBLE SOLUTION

M agpie S en sin g , a start-up p r o je c t u n d e r E bers Sm ith a n d D o u g las A sso ciate d LLC, p ro ­ vid es a suite o f c o ld ch a in m o n ito rin g an d an alysis te c h n o lo g ie s fo r th e h e a lth ca re in d u s­ try. It is a sh ip p a b le , w ire le s s tem p eratu re an d hum idity m o n ito r th a t p ro v id es real-tim e, lo ca tio n -a w a re tra ck in g o f c o ld ch a in p ro d u cts during sh ip m en t. M agpie S e n sin g ’s s o lu ­ tio n s re ly o n rich an alytics algorithm s th at le v e ra g e th e data g ath e re d fro m th e m o n ito r­ ing d e v ice s to im p ro ve th e e ffic ie n c y o f c o ld ch ain p ro c e s s e s a n d p re d ict c o ld sto rag e p ro b lem s b e fo r e th e y o ccu r.

M agp ie se n sin g a p p lie s all th ree ty p e s o f analytical te ch n iq u e s — d escrip tiv e, p re d ic­ tive, a n d p rescrip tiv e an aly tics— to tu rn th e raw data retu rn ed fro m th e m o n ito rin g d ev ice s into a c tio n a b le re co m m e n d a tio n s and w arnings.

T h e p ro p e rtie s o f th e c o ld s to ra g e sy stem , w h ic h in clu d e th e s e t p o in t o f th e sto rag e sy stem ’s th erm o stat, th e ty p ica l ran g e o f te m p e ratu re v a lu e s in th e s to ra g e system , a n d

3 4 Part I • D ecisio n M aking and Analytics: An Overview

the d uty c y c le o f th e sy stem ’s co m p re sso r, a re m o n ito re d an d re p o rte d in real tim e. This in form ation h e lp s train ed p e rs o n n e l to e n su re th at th e sto rag e unit is p ro p erly co n fig u re d to sto re a p articu lar p ro d u ct. All th e tem p eratu re in fo rm atio n is d isp layed o n a W e b d a sh ­ b o a rd th at s h o w s a g rap h o f th e tem p e ratu re in sid e th e s p e c ific s to ra g e unit.

B a s e d o n in form ation d eriv ed fro m th e m o n ito rin g d ev ice s, M ag p ie ’s p red ictiv e a n a ­ lytic algorithm s c a n d eterm in e th e s e t p o in t o f th e s to ra g e u n it’s th erm o stat a n d alert the sy stem s users i f th e s y stem is in co rre ctly co n fig u re d , d e p e n d in g u p o n th e vario u s typ es o f p ro d u cts stored . T h is o ffe rs a so lu tio n to th e u sers o f c o n s u m e r refrigerators w h e re th e th e rm o stat is n o t tem p e ratu re grad ed . M ag p ie ’s sy stem a lso s en d s alerts a b o u t p o s ­ sib le tem p eratu re v iolation s b a s e d o n th e s to ra g e u n it’s av e ra g e tem p eratu re and s u b s e ­ q u e n t c o m p re ss o r c y c le ru ns, w h ich m ay d rop th e te m p e ra tu re b e lo w th e fre e z in g point. M agpie s p red ictiv e analytics fu rth er re p o rt p o ss ib le h u m an errors, s u ch as failu re to shut th e sto ra g e unit d oors o r th e p re s e n c e o f an in c o m p le te se a l, b y analy zing th e te m p e ra ­ ture trend an d alertin g u sers via W e b in te rface , te x t m e s s a g e , o r au d ib le alert b e fo re th e tem p eratu re b o u n d s are actu ally violated . In a sim ilar w ay, a c o m p re ss o r o r a p o w e r failure c a n b e d etecte d ; th e e stim ated tim e b e fo re th e s to ra g e unit re a ch e s a n u n safe tem ­ p eratu re a lso is re p o rted , w h ich p re p a re s th e users to lo o k fo r b a c k u p so lu tio n s s u ch as u sin g dry ic e to resto re p o w er.

In ad dition to p red ictiv e analy tics, M agpie S e n s in g ’s an alytics sy stem s c a n p ro v id e p re scrip tiv e re co m m en d a tio n s fo r im p ro v in g th e c o ld sto rag e p ro c e s s e s an d b u sin ess d e cisio n m akin g. P rescriptive an aly tics h e lp u sers dial in th e op tim al tem p eratu re setting, w h ich h e lp s to a ch ie v e th e right b a la n c e b e tw e e n fre e z in g a n d s p o ila g e risk; this, in turn, p ro v id e s a cu sh io n -tim e to re a c t to th e situ ation b e fo re th e p ro d u cts sp o il. Its prescrip tiv e an alytics a lso g a th e r usefu l m eta-in fo rm atio n o n c o ld s to ra g e units, in clu d in g th e tim es o f d ay th a t a re b u s ie s t a n d p e rio d s w h e re th e sy stem ’s d o o rs a re o p e n e d , w h ic h c a n b e u se d to p ro v id e ad d itional d esig n p lan s an d institu tional p o lic ie s that e n su re that th e sy stem is b e in g p ro p e rly m ain tain ed an d n o t o v eru sed .

F u rth erm ore, p rescrip tive an alytics c a n b e u se d t o g u id e e q u ip m e n t p u rch a s e d e ci­ sio n s b y co n stan tly analyzing th e p e rfo rm a n ce o f cu rre n t s to ra g e units. B a sed o n th e s to ra g e sy stem ’s e fficie n cy , d ecisio n s o n d istributing th e p ro d u cts a cro ss a v ailab le storag e u nits c a n b e m ad e b a s e d o n th e p ro d u ct’s sensitivity.

U sing M ag p ie S e n sin g ’s c o ld ch a in analytics, ad d ition al m an u factu rin g tim e an d e x p e n d itu re c a n b e elim in ated b y e n su rin g that p ro d u ct safe ty c a n b e se cu re d th rou g h o u t th e su p p ly ch a in an d e ffe ctiv e p ro d u cts c a n b e ad m in istered to th e patients. C o m p lian ce w ith state and fed eral safe ty regu lation s c a n b e b etter a ch ie v e d th ro u g h au to m atic data g ath erin g an d re p o rtin g a b o u t th e p ro d u cts involved in th e co ld ch ain .

QUESTIONS FO R THE OPENING VIGNETTE

1 . W h at in form ation is pro v id ed b y th e d escrip tiv e an aly tics e m p lo y e d at M agpie Sensing?

2 . W h a t ty p e o f su p p o rt is pro v id ed b y th e p red ictiv e an aly tics e m p lo y e d at M agpie Sensing?

3 . H o w d o e s p rescrip tiv e analytics h e lp in b u sin ess d e c is io n m aking?

4 . In w h a t w ays c a n a c tio n a b le in form ation b e re p o r te d in re al tim e to c o n c e rn e d u sers o f th e system ?

5 . In w h at o th e r situ ation s m ight re al-tim e m o n ito rin g a p p lica tio n s b e need ed ?

WHAT WE CAN LEARN FROM THIS VIGNETTE

I his v ig n ette illustrates h o w data fro m a b u sin e ss p r o c e s s c a n b e u se d to g e n era te insights at various levels. First, th e grap h ical analysis o f th e data (te rm ed reportin g an aly tics) allow s

Chapter 1 • An O verview o f B usiness Intelligen ce, Analytics, and D ecision Support

users to g e t a g o o d fe e l fo r th e situation. T h e n , additional analysis u sin g data m ining .e ch n iq u es c a n b e u s e d to e stim ate w h at future b eh a v io r w o u ld b e like. T h is is th e d om ain o f pred ictive analytics. S u ch analysis c a n th e n b e ta k e n to cre ate sp e cific re co m m en d atio n s ror op erators. T h is is a n e x a m p le o f w h at w e call prescriptive analytics. Finally, this o p e n ­ ing vignette a lso suggests th at innovative ap p licatio n s o f analytics ca n c re a te n e w b u sin ess ventures. Id entifying op p ortu nities fo r ap p licatio n s o f analytics an d assisting w ith d ecisio n m aking in s p e cific d om ains is an e m e rg in g en trep ren eu rial opportunity.

Sources: Magpiesensing.com, "Magpie Sensing Cold Chain Analytics and Monitoring," magpiesensing.com/ wp-content:/upload.s/2013/01/ColdChainAnalyticsMagpieSensing-W hitepaper.pdf (accessed July 2013); Centers for Disease Control and Prevention, Vaccine Storage and Handling, http://www.cdc.gov/vaccines/pubs/ pinkbook/vac-storage.htmI*storage (accessed July 2013); A. Zaleski, “Magpie Analytics System Tracks Cold- -iafti Products to Keep Vaccines, Reagents Fresh ’ (2012). technicallybaltimore.com/profiles/startups/magpie- analytics-system-tracks-cold-chain-products-to-keep-vaccines-reagents-fresh (accessed February 2013).

1.2 CH AN G IN G BU SIN ESS ENVIRO NM ENTS A N D COM PUTERIZED D ECISIO N SUPPORT

T h e o p e n in g v ig n e tte illustrates h o w a co m p a n y can e m p lo y te c h n o lo g ie s to m a k e s e n se o f data a n d m a k e b etter d ecisio n s. C o m p an ie s are m o vin g aggressively to co m p u terized supp ort o f th e ir o p era tio n s. T o u n d erstan d w h y c o m p a n ie s are e m b ra cin g co m p u te r­ ized supp ort, in clu d in g b u s in e s s in te llig e n ce , w e d e v e lo p e d a m o d el c a lle d th e B u sin ess P ressu res-R espon ses-S u pport M odel, w h ich is sh o w n in Figure 1.1.

The Business Pressures-Responses-Support M odel T h e B u s in e s s P r e s s u r e s - R e s p o n s e s -S u p p o r t M o d el, as its n a m e in d ica te s , h a s th re e c o m ­ p o n e n ts: b u s in e s s p re s s u re s that resu lt fro m to d a y ’s b u s in e s s clim a te, re s p o n s e s (a c tio n s ta k e n ) b y c o m p a n ie s to c o u n te r th e p re ssu re s (o r to ta k e ad v a n ta g e o f th e o p p o rtu n itie s av ailab le in th e e n v iro n m e n t), an d c o m p u te riz e d su p p o rt th a t fa cilita te s th e m o n ito rin g o f th e e n v iro n m e n t an d e n h a n c e s th e r e s p o n s e a c tio n s ta k e n b y o rg a n iz a tio n s.

Decisions and Su p p o rt

FIGURE 1.1 The Business Pressures-Responses-Support Model.

3 6 P a r t i • D ecision M aking and Analytics: An Overview

5SS5 i ^ s S * = ^ K SS sssa&rss* •“*" T ^ iE is s E * * * * / . T h e s e ca te g o rie s a r i s ^ S b ^ " 0f i K * “ *

S w i S a ^ s s s Ltd (Krivda, 2 0 0 8 ), fo r e x a m p le tu rned to B l T ^ 6 preSSUreS' V o d a fo n e N ew Z ealand

E m p lo y strategic planning. • U se n e w and in n ov ativ e b u sin e ss m od els.

R estructu re b u sin ess p ro ce sse s. P articip ate in b u sin ess alliances.

• Im p ro v e co rp o ra te in form ation system s. • Im p ro v e p artn ersh ip relatio nsh ip s.

J A B L E 1.1 B usiness Environm ent Factors That Create Pressures on O rganizations

Factor __________ Description ~ "---- “ — —

Markets Strong competition Expanding global markets Booming electronic markets on the Internet Innovative marketing methods Opportunities for outsourcing with IT support Need for real-time, on-demand transactions

Consumer demands Desire for customization

Desire for quality, diversity of products, and speed of delivery Customers getting powerful and less loyal

Technology More innovationS/ new productSi and new servjcfis

Increasing obsolescence rate Increasing information overload Social networking, W eb 2.0 and beyond

Growing government regulations and deregulation Workforce more diversified, older, and composed of more women Prime concerns of homeland security and terrorist attacks Necessity of Sarbanes-Oxley Act and other reporting-related legislation Increasing social responsibility of companies Greater emphasis on sustainability

Societal

• E n co u ra g e in n o v atio n a n d creativity. • Im p rove cu sto m e r serv ice an d re latio n sh ip s. • E m p loy s o c ia l m e d ia a n d m o b ile platform s for e -c o m m e r c e and b ey o n d . • M o ve to m a k e -to -o rd e r p ro d u ctio n an d o n -d e m a n d m an u factu rin g a n d serv ices. • U se n e w IT to im p ro v e co m m u n ica tio n , data a c c e s s (d isco v e ry o f in fo rm atio n ), an d

c o lla b o ra tio n . • R e sp o n d q u ick ly to c o m p etito rs’ a ctio n s (e .g ., in pricing, p ro m o tio n s, n e w p ro d u cts

an d se rv ice s). • A u tom ate m an y task s o f w h ite -co lla r e m p lo y e e s. • A u tom ate ce rta in d e c is io n p ro ce s s e s, e sp ecia lly th o se d ealin g w ith cu sto m ers. • Im p ro v e d e c is io n m ak in g b y e m p lo y in g analytics.

Many, if n o t all, o f th e s e actio n s req u ire s o m e co m p u te rize d su p p ort. T h e s e a n d o th e r re sp o n se a c tio n s are fre q u e n tly facilitated b y co m p u te riz e d d e c is io n su p p o rt (D S S ).

C L O S IN G T H E S T R A T E G Y G A P O n e o f th e m a jo r o b je c tiv e s o f co m p u te riz e d d e c is io n su p p o rt is to fa cilita te c lo s in g th e gap b e tw e e n th e cu rre n t p e rfo rm a n c e o f an o rg an i­ z a tio n a n d its d e sire d p e rfo rm a n ce , as e x p r e s s e d in its m issio n , o b je c tiv e s , a n d goals, an d th e strate g y to a c h ie v e th e m . In o rd e r to u n d e rstan d w h y c o m p u te riz e d su p p o rt is n e e d e d a n d h o w it is p ro v id e d , e sp e c ia lly fo r d e cis io n -m a k in g su p p o rt, le t’s lo o k at m an ag erial d e c is io n m akin g.

SECTION 1 . 2 REVIEW QUESTIONS

1 . List th e c o m p o n e n ts o f an d e x p la in th e B u s in e s s P r e s s u r e s -R e s p o n s e s -S u p p o r t

M odel. 2 . W hat are s o m e o f th e m a jo r facto rs in to d ay ’s b u sin e ss environm ent?

3 . W h at a re s o m e o f th e m a jo r re s p o n s e activities that o rg an ization s take?

1.3 M A N A G E R IA L D ECISIO N M AKING M a n a g e m e n t is a p r o c e s s b y w h ic h o r g a n iz a tio n a l g o a ls a re a c h ie v e d b y u sin g re s o u r c e s . T h e r e s o u r c e s a re c o n s id e r e d in p u ts, a n d atta in m e n t o f g o a ls is v ie w e d as th e o u tp u t o f th e p r o c e s s . T h e d e g r e e o f s u c c e s s o f th e o r g a n iz a tio n a n d th e m a n a g e r is o fte n m e a s u re d b y th e ratio o f o u tp u ts to in p u ts. T h is ra tio is a n in d ic a tio n o f th e o r g a n iz a tio n ’s p ro d u ctiv ity , w h ic h is a r e fle c tio n o f t h e o r g a n iz a tio n a l a n d m a n a g e r ia l p er fo r m a n c e .

T h e le v e l o f pro d u ctiv ity o r th e s u c c e s s o f m a n a g e m e n t d e p e n d s o n th e p e rfo r­ m a n c e o f m a n a g e ria l fu n ctio n s, s u c h a s p la n n in g , o rg a n iz in g , d ire ctin g , a n d c o n tro l­ ling. T o p e rfo rm th e ir fu n c tio n s , m a n a g e rs e n g a g e in a c o n tin u o u s p r o c e s s o f m ak in g d e c is io n s . M a k in g a d e c is io n m e a n s s e le c tin g th e b e s t a ltern a tiv e fro m tw o o r m o re

so lu tio n s.

The Nature o f M anagers' W ork M intzberg’s (2 0 0 8 ) c la ss ic study o f to p m an ag e rs a n d sev eral re p lica te d stu d ies suggest that m a n a g e rs p e rfo rm 10 m a jo r ro le s that c a n b e classified into th ree m a jo r ca te g o iie s . in terperson al, in fo rm a tio n a l, an d d e c is io n a l (.s e e l a b l e 1.2).

T o p e rfo rm th e s e ro le s, m an ag e rs n e e d in fo rm atio n th at is d eliv e red efficie n tly an d in a tim ely m a n n e r to p e rso n a l co m p u te rs (P C s) o n th eir d esk to p s and to m o b ile d ev ices. T h is in form ation is d eliv e red b y n etw o rk s, g e n erally v ia W e b te ch n o lo g ie s.

In ad d itio n to o b ta in in g inform ation n e c e s s a ry to b e tte r p erfo rm th e ir ro le s, m an ag ­ ers u se c o m p u te rs d irectly to su p p o rt and im p ro ve d e c is io n m akin g, w h ic h is a k e y task

C hapter 1 • An O verview o f B u sin ess Intelligence, Analytics, and D ecision Support 3 7

D ecisio n M aking and Analytics: An Overview

T A B L E 1.2 M intzberg's 10 M anagerial Roles

Role Description Interpersonal Figurehead Is symbolic head; obliged to perform a number of routine duties of a

legal or social nature Leader Is responsible for the motivation and activation of subordinates;

responsible for staffing, training, and associated duties Liaison Maintains self-developed network of outside contacts and informers

who provide favors and information Informational Monitor Seeks and receives a wide variety of special information (much of it

current) to develop a thorough understanding of the organization and environment; emerges as the nerve center of the organization's internal and external information

Disseminator Transmits information received from outsiders or from subordinates to members of the organization; some of this information is factual, and some involves interpretation and integration

Spokesperson Transmits information to outsiders about the organization's plans, policies, actions, results, and so forth; serves as an expert on the organization's industry

Decisional Entrepreneur Searches the organization and its environment for opportunities and

initiates improvement projects to bring about change; supervises design of certain projects

Disturbance handler Is responsible for corrective action when the organization faces important, unexpected disturbances

Resource allocator Is responsible for the allocation of organizational resources of all kinds; in effect, is responsible for the making or approval of all significant organizational decisions

Negotiator Is responsible for representing the organization at major negotiations

Sources: Compiled from H. A. Mintzberg, The Nature o f M an agerial Work. Prentice Hall, Englewood Cliffs, NJ, 1980; and H. A. Mintzberg, The Rise a n d Fall o f Strategic P lanning. The Free Press, New York, 1993.

that is p art o f m o st o f th e s e ro le s. M any m an ag erial activities in all ro les rev o lv e around d e cisio n m akin g. M an agers, esp ecia lly th ose a t h ig h m a n a g e r ia l levels, a r e p rim a rily d e c i­ sio n m akers. W e re v iew th e d ecisio n -m a k in g p ro c e s s n e x t b u t w ill stud y it in m o re detail in th e n e x t ch ap ter.

The Decision-Making Process F o r years, m an ag ers co n sid e re d d e cisio n m ak in g p u re ly a n art— a talen t a cq u ire d o v e r a lo n g p e rio d th ro u g h e x p e rie n c e (i.e ., learn in g b y trial-an d -error) an d b y u sin g intuition. M an ag e m e n t w a s c o n s id e re d a n art b e c a u s e a v ariety o f individual styles c o u ld b e u sed in a p p ro a ch in g a n d s u cce ssfu lly so lv in g th e sa m e ty p e s o f m an ag erial p ro b lem s. T h e s e styles w e r e o fte n b a s e d o n creativity, ju d g m en t, intu ition, an d e x p e r ie n c e rath er than o n sy stem atic q u an titative m e th o d s g ro u n d e d in a s c ie n tific ap p ro a ch . H o w ev er, re ce n t re se a rch su g g e sts that c o m p a n ie s w ith to p m an ag e rs w h o a re m o re fo c u s e d o n p ersisten t w o rk (a lm o st d u lln ess) te n d to o u tp e rfo rm th o se w ith lead e rs w h o s e m a in strengths are in terp erso n al co m m u n icatio n skills (K a p la n e t al., 2 0 0 8 ; B ro o k s, 2 0 0 9 ). It is m o re im por­ tant to e m p h a s iz e m e th o d ical, thoughtful, an aly tical d e c is io n m ak in g rath er th an flashi­ n e s s a n d in terp erso n al co m m u n ica tio n skills.

Chapter 1 • An O verview o f B u sin ess Intelligence, Analytics, and D ecisio n Support 3 9

M an agers u su ally m a k e d e cis io n s b y fo llo w in g a fo u r-step p r o c e s s (w e le a rn m o re a b o u t th e s e in C h ap ter 2):

1 . D e fin e th e p ro b le m (i.e ., a d e c is io n situ atio n that m ay d ea l w ith s o m e d ifficulty o r w ith a n o p p ortu n ity ).

2. C o n stru ct a m o d el th at d e s c rib e s th e real-w orld p ro b lem . 3- Id en tify p o ss ib le so lu tio n s to th e m o d e le d p ro b lem a n d e v alu ate th e solu tions. 4 . C o m p are , c h o o s e , and re co m m en d a p o ten tial solu tion to th e p ro b lem .

T o fo llo w this p ro c e s s , o n e m u st m a k e sure th at su fficien t altern ativ e so lu tio n s are b e in g co n s id e re d , that th e c o n s e q u e n c e s o f u sin g th e s e alternativ es c a n b e re a so n a b ly p red icted , a n d that co m p a riso n s are d o n e p ro p erly. H ow ever, th e en v iro n m en tal facto rs listed in T a b le 1.1 m a k e s u ch a n evalu ation p ro c e s s d ifficult fo r th e fo llo w in g reason s:

• T e c h n o lo g y , inform ation sy stem s, ad v an ce d s e a rch e n g in e s , an d g lo b a liz a tio n result in m o re a n d m o re alternatives fro m w h ich to c h o o s e .

• G o v e rn m e n t reg u latio n s a n d th e n e e d fo r co m p lia n ce , p o litical instability an d te r­ rorism , co m p etitio n , and ch a n g in g c o n s u m e r d em an d s p ro d u ce m o re u n certainty, m a k in g it m o re d ifficu lt to p re d ict c o n s e q u e n c e s and th e future.

• O th e r fa c to rs are th e n e e d to m a k e rap id d ecisio n s, th e fre q u e n t an d u n p re d ictab le c h a n g e s that m ak e trial-an d -error learn in g difficult, and th e p o ten tial co s ts o f m aking m istak es.

• T h e s e e n v iro n m en ts a re g ro w in g m o re c o m p le x e v e ry day. T h e re fo re , m ak in g d e ci­ sio n s to d a y is in d e e d a c o m p le x task.

B e c a u s e o f th e s e tren d s an d ch a n g e s, it is n early im p o ssib le to rely o n a trial-and- e rro r a p p r o a c h to m an ag e m e n t, e sp e cia lly fo r d e cis io n s fo r w h ich th e fa c to rs sh o w n in T a b le 1.1 a re stro n g in flu e n ces. M anagers m u st b e m o re so p h isticated ; th e y m ust u s e th e n e w to o ls a n d te c h n iq u e s o f th eir field s. M ost o f th o se to o ls an d te c h n iq u e s are d iscu ssed in this b o o k . U sing th e m to su p p ort d e c is io n m ak in g c a n b e e x tre m e ly rew ard in g in m ak in g e ffe ctiv e d e cisio n s. In th e fo llo w in g s e ctio n , w e lo o k a t w h y w e n e e d co m p u te r su p p ort a n d h o w it is provided.

SECTION 1 . 3 REVIEW QUESTIONS

1 . D escrib e th e th ree m ajor m anagerial roles, and list som e o f the sp ecific activities in each.

2. W h y h a v e s o m e argu ed that m a n a g e m e n t is th e sam e as d e cisio n m aking? 3. D e s c r ib e th e fo u r ste p s m an ag ers ta k e in m ak in g a d ecisio n .

1.4 INFORM ATION S Y S T E M S SUPPORT FOR D ECISIO N M AKIN G From trad itional u s e s in p ayro ll a n d b o o k k e e p in g fu n ctio n s, co m p u te riz e d system s hav e p en e trate d c o m p le x m an agerial areas ran gin g fro m th e d esig n an d m an a g e m e n t o f au to ­ m ated fa cto rie s to th e a p p lica tio n o f analytical m eth o d s fo r th e e v a lu a tio n o f p ro p o s e d m ergers a n d acq u isitio n s. N early all e x e c u tiv e s k n o w that in form ation te c h n o lo g y is vital to th eir b u s in e s s an d e x te n siv e ly u se in form ation te ch n o lo g ie s.

C o m p u te r a p p lica tio n s h av e m o v e d fro m tran saction p ro ce s s in g a n d m o n ito rin g activities to p ro b le m analysis an d so lu tio n a p p licatio n s, an d m u ch o f th e activity is d o n e w ith W e b -b a s e d te ch n o lo g ie s, in m a n y c a s e s a c c e ss e d th ro u g h m o b ile d ev ice s. Analytics an d B I to o ls s u ch as data w areh o u sin g , d ata m ining, o n lin e analy tical p ro c e s s in g (O LAP), dashboards, an d th e u se o f th e W e b for d e cisio n su p p o rt a re th e co r n e r sto n e s o f to d ay ’s m o d e rn m an ag e m e n t. M anagers m ust h av e h ig h -sp e ed , n e tw o rk e d in form ation sys­ tem s (w ire lin e o r w ire le s s ) to assist th e m w ith th e ir m o st im p ortan t task : m ak in g d e ci­ sions. B e s id e s th e o b v io u s gro w th in h ard w are, so ftw are, a n d n e tw o rk c a p a c itie s , so m e

4 0 Part I • D ecisio n M aking and Analytics: An Overview

d ev elo p m en ts h a v e clearly co n trib u ted to facilitatin g gro w th o f d e c is io n su p p o rt and analytics in a n u m b e r o f w ay s, in clu d in g th e fo llow in g :

• G ro u p co m m u n ica tio n a n d collaboration. M any d ecisio ns are m ade to d ay b y groups w h o se m em b ers m ay b e in different locations. G roups ca n collab orate and com m u nicate readily b y using W e b -b a sed tools a s w e ll as th e ubiqu itous sm artphones. C ollaboration is esp ecially im portant alo n g th e supp ly chain, w h ere partners— all the w ay from vend ors to custom ers— m ust sh are inform ation. A ssem bling a group o f d ecisio n m akers, esp ecially experts, in o n e p la ce c a n b e costly. Infom iation system s c a n im prove th e collab oratio n p ro cess o f a gro u p an d e n a b le its m em b ers to b e at dif­ ferent locations (saving travel costs). W e will study so m e applications in Chapter 12.

• Im p ro v e d d a ta m a n a g e m e n t. M any d e cis io n s in v o lv e c o m p le x co m p u tatio n s. D ata fo r th e s e c a n b e sto red in d ifferen t d a ta b a s e s an y w h ere in th e o rg an ization a n d e v e n p o ss ib ly a t W e b sites ou tsid e th e org an izatio n . T h e data m ay in clu d e text, sou n d , grap h ics, a n d v id eo , an d th e y c a n b e in d ifferent lan g u ag es. It m ay b e n e c e s ­ sary to transm it data q u ick ly fro m d istant lo ca tio n s. System s to d ay c a n se a rch , store, an d tran sm it n e e d e d d ata q u ick ly , e co n o m ica lly , s e cu re ly , an d transp arently.

• M a n a g in g g i a n t d a ta w a reh o u ses a n d B ig D a ta . Large d ata w a reh o u se s, like th e o n e s o p e ra te d b y W alm art, co n ta in te ra b y tes an d e v e n p e ta b y tes o f d ata. Sp ecial m e th o d s, in clu d in g p arallel com p u tin g , are a v ailab le to o rg an ize , se a rch , a n d m ine th e d ata. T h e co sts related to d ata w a reh o u sin g are d eclin in g . T e c h n o lo g ie s th at fall u n d e r th e b ro ad ca te g o ry o f B ig D ata h av e e n a b le d m assiv e data c o m in g fro m a variety o f so u rce s and in m an y d ifferen t fo rm s, w h ich allo w s a v e ry d ifferen t v iew in to organ izatio n al p e rfo rm a n ce th at w as n o t p o ss ib le in th e past.

• A n a ly tic a l su p p o rt. W ith m o re data a n d a n aly sis te c h n o lo g ie s , m o re altern a­ tiv e s c a n b e e v alu ated , fo reca sts c a n b e im p ro v e d , risk an alysis c a n b e p e rfo rm e d q u ick ly , a n d th e v iew s o f e x p e rts (s o m e o f w h o m m ay b e in re m o te lo c a tio n s ) ca n b e c o lle c te d q u ick ly an d a t a re d u c e d co st. E x p e rtis e c a n e v e n b e d eriv ed d irectly fro m a n aly tical sy stem s. W ith s u ch to o ls , d e c is io n m a k e rs ca n p e rfo rm c o m p le x sim u latio n s, c h e c k m a n y p o ss ib le sce n a rio s, a n d a s s e s s d iv erse im p acts q u ic k ly and e c o n o m ic a lly . T h is, o f co u rse , is th e fo c u s o f s e v e r a l ch a p te rs in th e b o o k .

• O vercom ing cognitive lim its in p ro c e s s in g a n d sto rin g in fo rm a tio n . A ccording to Sim on (1 9 7 7 ), th e hu m an mind has only a lim ited ability to p ro cess and store infor­ mation. P e o p le som etim es find it difficult to recall an d use inform ation in a n error-free fash ion due to their cognitive limits. T h e term cogn itive lim its indicates that a n indi­ vidual’s problem -solving capability is limited w h e n a w id e range o f diverse inform ation and k n ow led g e is required. Com puterized system s e n a b le p e o p le to o v erco m e their cognitive limits b y q uickly a ccessin g a n d p ro cessin g vast am ounts o f stored inform ation (s e e C hapter 2).

• K n o w led g e m a n a g e m e n t . O rg an izatio n s h a v e g ath e re d vast s to res o f in form a­ tio n a b o u t th e ir o w n o p e ra tio n s, cu sto m e rs, in te rn a l p ro c e d u re s , e m p lo y e e in te ra c­ tion s, a n d s o fo rth th ro u g h th e u n stru ctu red a n d stru ctu red c o m m u n ica tio n s tak in g p la c e am o n g th e v ario u s s ta k e h o ld e rs. K n o w le d g e m a n a g e m e n t sy stem s (KM S, C h ap te r 12) h av e b e c o m e s o u rc e s o f fo rm al an d in form al su p p o rt fo r d e c is io n m ak in g to m a n a g e rs, a lth o u g h s o m e tim e s th e y m ay n o t e v e n b e ca lle d KMS.

• A n y w h ere, a n y tim e su p p o rt. U sing w ire le s s te c h n o lo g y , m an ag e rs c a n a c c e s s in form ation an y tim e an d fro m an y p la ce , an aly ze and in terp ret it, an d co m m u n icate w ith th o s e involved . T h is p e rh a p s is th e b ig g e st c h a n g e that h a s o ccu rre d in th e last fe w years. T h e s p e e d at w h ic h in form ation n e e d s to b e p ro c e s s e d a n d co n v e rte d in to d ecisio n s h a s truly c h a n g e d e x p e c ta tio n s fo r b o th co n su m e rs a n d b u sin esses.

T h e s e an d o th e r capabilities h av e b e e n driving th e u s e o f com p u terized d ecisio n supp ort s in ce th e late 1960s, b u t esp ecially sin ce th e m id -1990s. T h e grow th o f m o b ile tech n o lo g ies,

Chapter 1 • An O verview o f B usiness Intelligence, Analytics, and D ecisio n Support 41

social m ed ia platform s, and analytical to o ls has e n a b led a m u ch h ig h er lev el o f inform ation system s su p p ort fo r m anagers. In th e n e x t sectio n s w e study a historical classification o f d e cisio n su p p ort tasks. T h is lead s us to b e introd u ced to d ecisio n su p p ort system s. W e will th e n study a n o v erview o f te ch n o lo g ie s that hav e b e e n b road ly referred to as b u sin ess intel­ lig e n ce . F ro m th ere w e w ill b ro ad en o u r h orizons to introd u ce v arious types o f analytics.

SECTION 1 .4 REVIEW QUESTIONS

1 . W h a t are s o m e o f th e k e y sy stem -o rie n te d trend s that h av e fo s te re d IS-su p p orted d e c is io n m ak in g to a n e w level?

2 . List s o m e cap ab ilitie s o f in form ation sy stem s that c a n facilitate m an ag erial d e cisio n m akin g.

3 . H o w c a n a co m p u te r h e lp o v e rco m e the co g n itiv e lim its o f hum ans?

1.5 A N E A R L Y F R A M E W O R K FO R C O M PU T ER IZ ED D EC ISIO N SU PPO R T

An e a rly fram ew o rk fo r co m p u teriz ed d e c is io n su p p o rt in clu d es sev eral m a jo r co n c e p ts th at are u s e d in fo rth co m in g se ctio n s an d ch ap ters o f this b o o k . G o rry a n d Scott-M orton cre a te d a n d u s e d this fram ew o rk in th e early 19 7 0 s, an d th e fram ew o rk th e n e v o lv ed into a n e w te c h n o lo g y ca lle d DSS.

The G orry and Scott-M orton Classical Fram ework G orry an d Scott-M o rton (1 9 7 1 ) p ro p o s e d a fram ew o rk th at is a 3 -b y -3 m atrix, as s h o w n in Figure 1.2 . T h e tw o d im en sio n s are th e d eg re e o f stru ctu red n ess a n d th e ty p e s o f con trol.

Type of Control

Type of Decision Operational

Control M anagerial

Control S tra te g ic Planning

S tru c tu re d Accounts receivable Accounts payable Order entry

Budget analysis Short-term forecasting Personnel reports Make-or-buy

Financial management Investment portfolio W arehouse location Distribution systems

Production scheduling Inventory control

Se m istru ctu re d

Credit evaluation Budget preparation Plant layout Project scheduling Reward system design Inventory

categorization

Building a new plant M ergers & acquisitions New product planning Compensation planning Quality assurance HR policies Inventory planning

8

U n stru ctu re d Buying software Approving loans Operating a help desk Selecting a cover for

a magazine

Negotiating Recruiting an executive Buying hardware Lobbying

R S . D planning New tech development Social responsibility

planning

FIGURE 1.2 Decision Support Frameworks.

4 2 Part I • D ec isio n M aking and Analytics: An O verview

D E G R EE OF STRU C TU RED N ESS T h e left sid e o f F ig u re 1 .2 is b a se d o n S im o n ’s (1 9 7 7 ) idea that d ecisio n -m a k in g p ro c e s s e s fall alo n g a co n tin u u m that ra n g e s fro m hig hly structured (s o m e tim e s called p rog ram m ed ) to h ig h ly u n stru ctu red (i.e ., n on p rog ram m ed ) d ecisio n s. Stru ctu red p ro c e s s e s a re ro u tin e and typ ically re p e titiv e p ro b lem s fo r w h ich standard so lu tio n m e th o d s exist. U n stm ctu red p ro cesses are fuzzy, c o m p le x p ro b lem s fo r w h ich th e re a re n o cu t-and -d ried so lu tio n m ethod s.

An u n stru ctu red problem is o n e w h e re th e articu latio n o f th e p ro b le m o r th e so lu ­ tio n a p p ro a ch m ay b e un stru ctu red in itself. In a stru ctu red problem , th e p ro ced u res fo r o b ta in in g th e b e s t (o r a t le a s t a g o o d e n o u g h ) so lu tio n a re k n o w n . W h e th e r th e p ro b ­ lem involves find ing an ap p rop riate inventory le v e l o r ch o o s in g a n op tim al in v estm en t strategy, th e o b je c tiv e s a re clearly d efin ed . C o m m o n o b je c tiv e s are c o s t m inim ization and p ro fit m axim ization.

Sem istructured problem s fall b e tw e e n stru ctu red a n d u n stru ctu red p ro b le m s, h av ­ ing s o m e stru ctu red e le m e n ts a n d so m e u n stru ctu red e le m e n ts. K e e n a n d Sco tt-M o ito n ( 1 9 7 8 ) m e n tio n e d trad ing b o n d s, settin g m arketin g b u d g ets fo r c o n s u m e r p ro d u cts, and p erfo rm in g cap ital a cq u isitio n analysis as sem istru ctu red p ro b lem s.

T Y P E S OF CONTROL T h e s e c o n d h a lf o f th e G o r ry an d S c o tt-M o rto n fra m ew o rk (r e fe r to F ig u re 1 .2 ) is b a s e d o n A n th o n y ’s ( 1 9 6 5 ) ta x o n o m y , w h ic h d e fin e s th re e b r o a d c a te g o r ie s th at e n c o m p a s s all m a n a g e ria l a c tiv itie s : stra teg ic p la n n in g , w h ic h in v o lv e s d e fin in g lo n g -r a n g e g o a ls a n d p o lic ie s fo r r e s o u r c e a llo c a tio n ; m a n a g e­ m en t co n tro l, th e a c q u is itio n a n d e ffic ie n t u s e o f r e s o u r c e s in th e a c c o m p lis h m e n t o f o r g a n iz a tio n a l g o a ls ; a n d o p e r a tio n a l co n tro l, th e e ffic ie n t an d e ffe c tiv e e x e c u t io n o f s p e c ific task s.

THE D EC ISIO N SU PPO RT M A T R IX A n th on y ’s an d S im o n ’s tax o n o m ie s are c o m b in e d in the n in e -c e ll d e c is io n su p p ort m atrix s h o w n in Figure 1.2 . T h e initial p u rp o s e o f this m atrix w a s to su g g e st d ifferen t typ es o f co m p u te rize d su p p o rt to d ifferent c e lls in th e matrix. G orry an d Scott-M o rton su gg ested , fo r e x a m p le , th a t fo r sem istru ctu red d ecisio n s and u n stru ctu red d ecisio n s, co n v e n tio n a l m a n a g e m e n t in form ation sy stem s (M IS) a n d m an­ a g e m e n t s c ie n c e (M S) to o ls are in su fficien t. H u m an in te lle ct a n d a d ifferen t a p p ro a ch to co m p u te r te c h n o lo g ie s a re n e cessa ry . T h e y p ro p o s e d th e u se o f a su p p ortiv e inform ation system , w h ic h th ey ca lle d a DSS.

N o te th a t th e m o re stru ctu re d a n d o p e r a tio n a l c o n tr o l-o r ie n te d ta s k s ( s u c h as th o s e in c e lls 1, 2, a n d 4 ) a re u su a lly p e rfo rm e d b y lo w e r-le v e l m a n a g e rs , w h e re a s th e ta s k s in c e lls 6 , 8 , an d 9 a re th e re s p o n s ib ility o f to p e x e c u tiv e s o r h ig h ly train e d s p e c ia lis ts .

Com puter Support fo r Structured Decisions C o m p u ters h av e historically su p p o rte d stru ctured an d s o m e sem istru ctu red d ecisio n s, e sp e cia lly th o s e th at involve o p era tio n a l an d m an ag erial co n tro l, s in c e th e 1960s. O p e ra tio n a l a n d m anag erial co n tro l d e cis io n s are m a d e in all fu n ctio n al areas, e sp ecia lly in fin a n ce an d p ro d u ctio n (i.e ., o p e ra tio n s) m an ag em en t.

S ta ic tu r e d p ro b le m s, w h ich a re e n c o u n te re d re p e a te d ly , h av e a h ig h le v e l o f stru c­ tu re. It is th e re fo r e p o s s ib le to ab stract, an aly ze, a n d classify th e m into s p e c ific c a te g o ­ ries. F o r e x a m p le , a m a k e -o r-b u y d e c is io n is o n e ca te g o ry . O th e r e x a m p le s o f c a te g o rie s a re cap ital b u d g etin g , a llo ca tio n o f re s o u r c e s , d istrib u tio n , p ro cu re m e n t, p lan n in g , and in v e n to ry co n tro l d e cisio n s. F o r e a c h c a te g o ry o f d e c is io n , a n e asy -to -a p p ly p re scrib e d m o d e l an d so lu tio n a p p r o a c h h a v e b e e n d e v e lo p e d , g e n e ra lly as q u antitative form ulas. T h e re fo re , it is p o ss ib le to u se a s c ie n tific a p p ro a c h fo r au to m atin g p o rtio n s o f m a n a g e ­ rial d e c is io n m akin g.

Chapter 1 • A n O verview o f B u sin ess Intelligen ce, Analytics, and D ecisio n Support 4 3

Com puter Support fo r Unstructured Decisions U nstructured p ro b le m s c a n b e o n ly partially su p p o rte d b y standard co m p u te riz e d q u a n ­ titative m e th o d s. It is u su ally n e ce s s a ry to d e v e lo p cu sto m ized solu tion s. H o w e v e r, su ch so lu tion s m ay b e n e fit fro m data a n d in form ation g e n era ted fro m c o rp o ra te o r e xte rn al data so u rce s. In tu itio n an d ju d g m en t m ay p lay a larg e ro le in th e s e ty p es o f d e cisio n s, as m ay co m p u te riz e d co m m u n icatio n a n d co lla b o ra tio n te ch n o lo g ie s, as w e ll a s k n o w le d g e m an ag e m e n t ( s e e C h ap ter 12).

Com puter Support fo r Sem istructured Problem s Solving sem istru ctu red p ro b lem s m ay in v o lv e a co m b in a tio n o f stand ard so lu tio n p ro­ ce d u re s and h u m a n ju d gm en t. M an ag em en t s c ie n c e c a n p ro v id e m o d els fo r th e p o rtio n o f a d e cisio n -m a k in g p ro b le m that is structured. F o r th e unstructured p o rtio n , a D SS c a n im prove th e q u ality o f th e in form ation o n w h ich th e d e cisio n is b a s e d b y p roviding, fo r e x a m p le , n o t o n ly a sin gle so lu tio n b u t a lso a ran g e o f alternative so lu tio n s, a lo n g w ith th eir p o ten tia l im pacts. T h e s e cap ab ilitie s h e lp m an ag e rs to b e tte r u n d e rstan d th e natu re o f p ro b lem s a n d , thu s, to m a k e b e tte r d ecisio n s.

SECTION 1 .5 REVIEW QUESTIONS

1 . W h at are stru ctu red , un structured , and sem istru ctu red d ecisio n s? P ro v id e tw o e x a m ­ p le s o f e a c h .

2. D e fin e o p era tio n a l con trol, m a n a g eria l con trol, an d strateg ic p la n n in g . P rov id e tw o e x a m p le s o f e a ch .

3. W h at are th e n in e c e lls o f th e d e cisio n fram ew ork? E x p la in w h at e a c h is for. 4 . H o w c a n co m p u te rs pro v id e su p p o rt fo r m ak in g stru ctu red decisions? 5. H o w c a n co m p u te rs pro v id e su p p o rt to sem istru ctu red a n d un stru ctu red decisions?

1.6 THE CON CEPT OF DECISIO N SUPPORT SYST EM S (D SS) In th e e a rly 1 9 7 0 s , Scott-M o rton first articu lated th e m a jo r c o n c e p ts o f D SS. H e d efin ed decision su p p o rt system s (DSS) as “in teractiv e co m p u te r-b a se d sy stem s, w h ich h elp d e cisio n m a k e rs u tilize d a ta an d m od els to so lv e unstructured p ro b le m s ” (G o rry and Scott-M orton , 1 9 7 1 ). T h e fo llo w in g is a n o th e r cla ssic D SS d efin ition , p ro v id e d b y K e en an d Sco tt-M o rto n (1 9 7 8 ):

D e c is io n su p p o rt system s c o u p le th e in tellectu al re so u rce s o f individ uals w ith th e cap a b ilitie s o f th e co m p u te r to im p ro ve th e qu ality o f d e cisio n s. It is a c o m p u te r-b a s e d su p p o rt sy stem fo r m an a g e m e n t d e c is io n m ak ers w h o deal w ith sem istru ctu red p ro b lem s.

N ote th a t th e term d ecisio n su pport system , lik e m an ag em en t in fo rm a tio n system (M IS) an d o th e r te rm s in th e field o f IT , is a c o n te n t-fre e e x p r e s s io n ( i.e ., it m e a n s d ifferent things to d iffe ren t p e o p le ). T h e re fo re , th e re is n o u n iv ersally a c c e p te d d efin itio n o f D SS. (W e p re s e n t ad d itio n al d efin itio n s in C h ap ter 2 .) Actually, D SS c a n b e v ie w e d a s a c o n ­ cep tu a l m eth od olog y — th at is, a b ro a d , u m b rella term . H o w ev er, s o m e v ie w D SS as a nar­ ro w er, s p e c ific d e c is io n su p p o rt ap p licatio n .

DSS as an Um brella Term T h e term DSS c a n b e u se d as a n u m brella term to d e s c rib e any co m p u te riz e d sy stem that su p p orts d e c is io n m ak in g in a n organization. An o rg an izatio n m ay h a v e a k n o w le d g e

4 4 Part I * D ec isio n M aking and Analytics: An O verview

m a n a g e m e n t sy stem to g u id e all its p e rs o n n e l in th eir p ro b le m solving. A n o th e r o rg an iza­ tio n m ay h av e sep arate su p p o rt system s fo r m arketing , fin a n ce , an d a cco u n tin g ; a su p ­ p ly ch a in m a n a g e m e n t (SC M ) sy stem fo r p ro d u ctio n ; a n d sev eral ru le -b a s e d sy stem s fo r p ro d u ct re p a ir d iag n o stics an d h e lp d esk s. D SS e n c o m p a s s e s th e m all.

Evolutio n of DSS into Business intelligence I n th e e a rly d ay s o f D SS, m a n a g e rs le t th e ir s ta ff d o s o m e s u p p o rtiv e a n a ly sis b y u sin g D SS to o ls . As PC te c h n o lo g y a d v a n c e d , a n e w g e n e r a tio n o f m a n a g e rs e v o lv e d — o n e th a t w a s c o m fo r ta b le w ith co m p u tin g a n d k n e w th a t te c h n o lo g y c a n d ir e c tly h e lp m a k e in te llig e n t b u s in e s s d e c is io n s fa ste r. N ew to o ls s u c h a s OLAP, d ata w a re h o u s in g , data m in in g , a n d in te llig e n t s y stem s, d e liv e re d v ia W e b te c h n o lo g y , a d d e d p ro m ise d ca p a b ilitie s a n d e a sy a c c e s s to to o ls , m o d e ls, a n d d a ta fo r c o m p u te r-a id e d d e c is io n m a k in g . T h e s e to o ls sta rte d to a p p e a r u n d e r th e n a m e s B I and b u sin ess a n a ly tic s in th e m id -1 9 9 0 s . W e in tro d u c e th e s e c o n c e p t s n e x t, a n d re la te th e D SS an d B I c o n c e p ts in th e fo llo w in g s e c tio n s .

SECTION 1 .6 REVIEW QUESTIONS

1 . P rov id e tw o d efin itio n s o f DSS. 2 . D e s c r ib e DSS a s an u m brella term .

1.7 A FRAM EW ORK FOR BU SIN ESS IN TELLIG EN C E (BI) T h e d e c is io n su p p o rt c o n c e p ts p re se n te d in S e ctio n s 1.5 an d 1.6 h av e b e e n im p lem en ted in crem en tally , u n d e r d ifferen t n am e s, b y m an y v e n d o rs th a t h av e c re a te d to o ls an d m e th ­ o d o lo g ie s fo r d e c is io n su p p ort. As th e e n te rp rise-w id e sy stem s grew , m an ag ers w e re a b le to a c c e s s u ser-frien d ly reports that e n a b le d th e m to m a k e d e cis io n s q u ick ly . T h e s e sy stem s, w h ich w e re g e n erally ca lle d ex ecu tiv e in fo rm a tio n system s (E IS ), th e n b e g a n to o ffe r ad d itional v isualization , alerts, an d p e rfo rm a n ce m e a su re m e n t cap ab ilitie s. B y 2 0 0 6 , th e m a jo r co m m erc ia l p ro d u cts an d s erv ices a p p e a re d u n d er th e u m brella term bu sin ess in tellig en ce (B I).

D efinitions o f BI B usiness intelligence (B I) is a n u m brella term th at c o m b in e s arch itectu res, to o ls, data­ b a se s , analy tical to o ls, a p p licatio n s, an d m e th o d o lo g ie s. It is, lik e D SS, a co n ten t-fre e e x p re s s io n , s o it m e a n s d ifferen t things to d ifferen t p e o p le . P art o f th e c o n fu sio n a b o u t B I lies in th e flurry o f a cro n y m s an d b u zzw o rd s th at a re a s s o cia te d w ith it (e .g ., b u sin e ss p e rfo rm a n ce m an a g e m e n t [BPM]). B I ’s m a jo r o b je c tiv e is to e n a b le in teractiv e a c c e ss (s o m e tim e s in re al tim e ) to data, to e n a b le m an ip u latio n o f d ata, an d to give b u sin ess m an ag ers and analysts th e ability to co n d u c t a p p ro p riate analyses. B y analyzing historical a n d cu rre n t d ata, situations, an d p e rfo rm a n ce s, d e c is io n m ak ers g e t v a lu a b le insights that e n a b le th e m to m ak e m o re in fo rm ed an d b e tte r d e c is io n s . T h e p ro c e s s o f B I is b ase d o n th e tran sform ation o f data to in form ation , th e n to d e cisio n s, an d finally to action s.

A Brief H istory o f BI T h e term B I w as c o in e d b y th e G artn er G ro u p in th e m id -1 9 9 0 s. H o w ev er, th e c o n c e p t is m u ch o ld er; it h a s its ro ots in th e MIS re p o rtin g sy stem s o f th e 1970s. D u ring th at p erio d , rep ortin g sy stem s w e re static, tw o d im en sio n al, a n d had n o analytical cap ab ilities. In the e a rly 1 9 8 0 s, th e c o n c e p t o f ex ecu tiv e in fo rm a tio n system s (E IS ) em e rg ed . T h is c o n c e p t e x p a n d e d th e co m p u te rize d su p p o rt to to p -lev el m a n a g e rs an d e x e cu tiv e s . S o m e o f th e

Chapter 1 • An O verview o f B usiness Intelligen ce, Analytics, and D ecision Support

R G U R E 1.3 Evolution of Business Intelligence (Bl).

cap ab ilities in tro d u ced w e re d y n am ic m u ltid im en sion al (a d h o c o r o n -d e m a n d ) reporting, fo reca stin g an d p re d ictio n , tre n d analysis, drill-d ow n to d etails, statu s a c c e ss , a n d criti­ c a l s u c c e s s facto rs. T h e s e fe a tu res a p p e a re d in d o z e n s o f co m m e rcia l p ro d u cts until the m id -1990s. T h e n th e sa m e cap ab ilitie s a n d s o m e n e w o n e s a p p e a re d u n d e r th e n a m e B I. 7 odav. a g o o d B l-b a s e d e n te rp rise inform ation sy stem co n ta in s all th e in fo rm atio n e x e c u ­ t e s n e ed . S o , th e orig in al c o n c e p t o f EIS w a s tran sform ed in to B I. B y 2 0 0 5 , B I system s s e in e d to in clu d e a r tific ia l in tellig en ce cap ab ilitie s as w e ll as p o w erfu l an aly tical ca p a b ili­ ties. Figu re 1 .3 illustrates th e vario u s to o ls and te c h n iq u e s that m ay b e in clu d e d in a B I svstem . It illustrates th e e v o lu tio n o f B I as w ell. T h e to o ls s h o w n in F ig u re 1.3 p ro v id e the cap ab ilities o f B I. T h e m o st so p h istica te d B I p ro d u cts in clu d e m o s t o f th e s e cap ab ilities; oth ers s p e cia liz e in o n ly s o m e o f th em . W e w ill stud y sev eral o f th e s e ca p a b ilitie s in m o re detail in C h ap ters 5 th ro u g h 9.

The Architecture o f BI A B I system has four m ajor com ponents: a d a ta w arehou se, w ith its so u rce data; business an alytics, a co llectio n o f tools fo r manipulating, m ining, and analyzing th e data in th e data w areho u se; business p erfo rm a n ce m an agem en t (BPM) fo r m onitoring a n d analyzing perfor­ m ance; and a u ser in terface (e .g ., a dashboard). T h e relationship am ong th e se com p on en ts is illustrated in Figure 1.4. W e will discuss th ese com p on en ts in detail in Chapters 3 through 9-

Styles o f BS T h e a rch ite ctu re o f B I d e p e n d s o n its a p p licatio n s. M icroStrategy C o rp . d istinguishes five styles o f B I a n d o ffers s p e c ia l to o ls fo r e a c h . T h e fiv e styles are re p o rt d elivery a n d alert­ ing; e n te rp rise re p o rtin g (u sin g d ash b o ard s and s c o re ca rd s ); c u b e an aly sis (a ls o k n o w n as s lice -a n d -d ice an alysis); ad h o c q u e ries; a n d statistics a n d data m ining.

4 6 Part I • D ec isio n M aking and Analytics: An Overview

FIGURE 1.4 A High-Level Architecture of Bl. Source: Based on W. Eckerson, Smart Companies in the 21st Century: The Secrets o f Creating Successful Business Intelligent Solutions. The Data Warehousing Institute, Seattle, WA, 2003, p. 32, Illustration 5.

The Origins and D rivers o f Bl W h e re did m o d e rn a p p ro a c h e s to data w a reh o u sin g (D W ) and B l c o m e from? W h at are th e ir ro ots, and h o w d o th o s e ro o ts a ffe c t t h e w a y o rg an izatio n s are m an agin g th e s e initia­ tives today? T o d a y ’s in v estm en ts in in form ation te c h n o lo g y a re u n d e r in cre ase d scrutiny in term s o f th e ir b o tto m -lin e im p act and p o ten tial. T h e sam e is tru e o f D W an d th e B l ap p lica tio n s that m a k e th e s e initiatives p o ssib le.

O rg an izatio n s are b e in g c o m p e lle d to cap tu re, un d erstan d , a n d h a rn e ss th eir data to su p p o rt d e c is io n m ak in g in ord e r to im p ro ve b u sin e ss o p era tio n s. L egislation an d regu lation (e .g ., th e S a rb a n e s -O x le y A ct o f 2 0 0 2 ) n o w req u ire b u sin e ss lead e rs to d o cu ­ m e n t th e ir b u sin e ss p ro c e s s e s a n d to sig n o f f o n th e leg itim acy o f th e in form ation th ey rely o n an d re p o rt to stak e h o ld e rs. M oreover, b u s in e s s c y c le tim es are n o w e xtrem ely co m p re sse d ; faster, m o re in form ed , and b e tte r d e c is io n m ak in g is th e re fo re a com p etitiv e im perative. M an agers n e e d the right in form ation at th e right tim e an d in th e right p la c e . T h is is th e m antra fo r m o d e rn a p p ro a ch e s to B l.

O rganizatio ns have to w o rk smart. P aying carefu l atten tion to th e m an ag e m e n t o f B l initiatives is a n e cessa ry a s p e c t o f d o in g b u sin ess. It is n o surprise, th en , that o ig an izatio n s are in creasin g ly ch am p io n in g B l. Y o u will h e a r ab ou t m o re B l s u c c e s s e s an d th e funda­ m entals o f th o se s u c c e s s e s in C hapters 3 th ro u g h 9. E x a m p le s o f m any ap p licatio n s o f B l are provid ed in T a b le 1.3. A p plication C ase 1.1 illustrates o n e s u ch a p p licatio n o f B l that has h e lp ed m any airlines, as w e ll as th e co m p a n ie s o ffe rin g s u ch serv ices to th e airlines.

A M ultim edia Exercise in Business Intelligence Terad ata University N etw ork (TU N ) inclu des so m e v id e o s alo n g th e lin es o f the televi­ sio n sh o w CSI to illustrate co n c e p ts o f analytics in d ifferent industries. T h e s e are called “B S I V id eo s (B u sin e ss Scen ario Investigations).” N o t o n ly th e se are entertaining, but th ey also p rovide th e class w ith so m e q u estio ns fo r d iscu ssion. F or starters, p lease g o to teradatauniversitynetwork.com/teach-and-leam/library-item/?LibraryItemId=889 W atch th e vid eo that a p p ea rs o n Y o u T u b e . Essentially, you have to assu m e th e ro le o f a cu stom er service ce n te r professional. An in com in g flight is running late, and several pas­ sen gers are likely to m iss th eir co n n ectin g flights. T h e re are seats o n o n e ou tg o in g flight that c a n acco m m o d ate tw o o f th e fo u r p assen gers. W h ich tw o passen gers sh ou ld b e given

C hapter 1 • An Overview o f B u sin ess Intelligen ce, Analytics, and D ecisio n Support 4 7

T A B L E 1.3 Business V alu e of BI A nalytical Applications u -11 i 11111 i | i 111§ 11;

Analytic Application Business Q uestion Business V alue

Customer segmentation W hat market segments do my customers fall into, and what are their characteristics?

Personalize customer relationships for higher satisfaction and retention.

Propensity to buy Which customers are most likely to respond to my promotion?

Target customers based on their need to increase their loyalty to your product line.

Also, increase campaign profitability by focusing on the most likely to buy.

Customer profitability W hat is the lifetime profitability of my customer?

Make individual business interaction decisions based on the overall profitability of customers.

Fraud detection How can I tell which transactions are likely to be fraudulent?

Quickly determine fraud and take immediate action to minimize cost.

Customer attrition Which customer is at risk of leaving? Prevent loss of high-value customers and let go of lower-value customers.

Channel optimization W hat is the best channel to reach my cus­ tomer in each segment?

Interact with customers based on their preference and your need to manage cost.

Source: A. Ziama and J. Kasher, D ata M ining Prim er f o r the D ata W arehousing Professional. Teradata, Dayton, OH, 2004.

Application Case 1.1 Sabre Helps Its Clients Through Dashboards and Analytics Sabre is o n e o f the w orld leaders in th e travel indus­ try, providing b o th b u sin ess-to -consu m er services as well as b u sin ess-to -bu siness services. It serves travel­ ers, travel agents, corporations, and travel suppliers through its fo u r m ain com p an ies: Travelocity, Sabre Travel N etw ork, Sab re Airline Solutions, and Sabre Hospitality Solutions. T h e cu rrent volatile global e c o ­ no m ic e nvironm ent p o se s significant com petitive chal­ leng es to th e airline industry. T o stay ah e a d o f the com p etition, Sabre Airline Solutions recog n ized that airline e x e cu tiv es n eed ed en h a n ce d to o ls fo r m anag­ ing their b u sin ess d ecisio ns b y elim inating the tradi­ tional, m anual, tim e-consu m ing p ro cess o f co llect­ ing an d aggregating financial and o th er inform ation n eed ed fo r action ab le initiatives. This e n a b les real-tim e d ecisio n .support at airlines throughout th e w orld that m axim ize their (and, in turn, S a b re ’s) return o n infor­ m ation b y driving insights, action ab le intelligence, and value fo r cu stom ers fro m the grow ing data.

Sab re d ev elo p e d a n Enterprise Travel D ata W are h o u se (E T D W ) using Terad ata to hold its m as­ siv e reservations data. E TD W is up dated in near-real tim e w ith b a tch es that run e v ery 15 m inutes, gathering

data fro m all o f S ab re ’s b u sin esses. S ab re u se s its ETD W to cre a te Sab re E xecu tiv e D ashbo ard s that p ro ­ vid e n e a r-re a l-tim e e x e cu tiv e insights using a C ognos 8 B I platform w ith O racle D ata Integrator an d O ra cle G old en gate te ch n o lo g y infrastructure. T h e E xecutive D ash bo ard s o ffe r their clie n t airlines’ top-lev el m an­ agers and d e cisio n m ak ers a tim ely, autom ated , user- friendly solu tion, aggregating critical perfo rm an ce m etrics in a su ccin ct w ay an d providing at a glance a 3 6 0 -d e g ree v ie w o f the overall h ealth o f th e airline. At o n e airline, Sab re’s E xecu tiv e D ash b o ard s provide sen io r m an ag e m e n t w ith a daily and intra-day sn a p ­ sh o t o f k e y p erfo rm an ce indicators in a sin gle appli­ cation, re p lacin g th e o n ce -a -w e e k , 8-h o u r p ro cess o f generating th e sam e rep o rt fro m various data sou rces. T h e use o f d ash bo ard s is n o t lim ited to th e external cu stom ers; S ab re also u se s th e m fo r th eir assessm en t o f internal o p eratio n al perform ance.

T h e dashboards h elp Sab re's cu stom ers to have a clea r understanding o f th e data through d ie visual displays that in corp o rate interactive drill-dow n cap a­ bilities. It rep laces flat presentations and allow s for m o re fo cu se d review o f th e data w ith less effort and

('C on tin u ed )

4 8 Part I • D ecisio n M aking and Analytics: An Overview

Application Case 1.1 (Continued) tim e. This facilitates team d ialog b y m aking the data/ m etrics p ertainin g to sales p erform ance, including ticketing, seats sold a n d flow n, op erational perfor­ m a n ce s u ch as data o n flight m o v em en t and track­ ing, cu stom er reservations, inventory, and revenue acro ss a n airline’s multiple distribution channels, avail­ a b le to m any stakeholders. T h e dashboard system s provide scalab le infrastructure, graphical u se r interface (G U I) supp ort, data integration, an d data aggregation that e m p o w e r airline execu tiv es to b e m o re proactive in taking action s that lead to positive im pacts o n the overall health o f th eir airline.

W ith its ETD W , Sab re co u ld also d ev elo p o th er W e b -b a s e d analytical an d reporting solutions that lev ­ era g e data to g ain cu sto m e r insights through analysis Of cu sto m e r p ro files an d th eir s ales interactions to ca l­ cu late cu sto m e r value. T h is e n a b les b etter cu stom er seg m en tatio n a n d insights fo r v alu e-ad d ed services.

Q u e s t i o n s f o r D i s c u s s i o n

1. W h at is trad itional reporting? H o w is it u se d in organizations?

2. H o w c a n an aly tics b e u sed to tran sform tradi­ tion al reporting?

3- H o w c a n in teractiv e rep ortin g assist o rg an iza­ tio n s in d e c is io n m aking?

W h a t W e C a n L e a m f r o m T h is A p p lic a tio n C a se

T h is A p p lic a tio n C a s e s h o w s th at o rg a n iz a tio n s th a t e a r lie r u s e d re p o r tin g o n ly fo r tra c k in g th e ir in te rn a l b u s in e s s activ itie s a n d m e e tin g c o m p lia n c e r e q u ire m e n ts s e t o u t b y th e g o v e r n m e n t a re n o w m o v in g to w ard g e n e r a tin g a c tio n a b le in te llig e n c e fro m th e ir tra n s a c tio n a l b u s in e s s data. R e p o rtin g h a s b e c o m e b r o a d e r a s o r g a n iz a tio n s a re n o w try­ in g to a n a ly z e a rc h iv e d tra n sa c tio n a l d ata to u n d e r­ sta n d u n d e rly in g h id d e n tre n d s a n d p a tte rn s that w o u ld e n a b le th e m to m a k e b e tt e r d e c is io n s b y g a in in g in sig h ts in to p ro b le m a tic a re a s a n d re s o lv ­ in g th e m to p u rs u e cu rre n t an d fu tu re m a rk e t o p p o rtu n itie s . R e p o rtin g h a s a d v a n c e d to in te ra c ­ tiv e o n lin e r e p o r ts th a t e n a b le u s e rs t o pu ll and q u ic k ly b u ild c u s to m re p o r ts as re q u ire d a n d e v e n p re s e n t th e re p o r ts a id e d b y v is u a liz a tio n to o ls th a t h a v e th e a b ility to c o n n e c t to th e d a ta b a s e , p ro v id in g th e c a p a b ilitie s o f d ig g in g d e e p into su m m a riz ed d ata.

Source: Teradata.com, “Sabre Airline Solutions/'teradata.eom/t/ case-stu d ies/ Sabre-A ir!ine-Solution s-E B 6281 (accessed February 2013).

priority? Y o u are given inform ation a b o u t cu sto m ers’ profiles an d relationsh ip w ith th e air­ line. Y o u r d ecisio n s m ight ch a n g e as you learn m o re a b o u t th o se cu stom ers’ profiles.

W atch th e v id eo , p au se it as ap p ro p riate, an d a n s w e r th e q u e stio n s o n w h ich p as­ se n g e rs sh o u ld b e g iv e n priority. I h e n re su m e th e v id e o to g e t m o re in form ation . After th e v id e o is c o m p le te , y o u c a n s e e th e slid es re la te d to this v id eo and h o w th e analysis w a s p re p a re d o n a slid e s e t at teradatauniversitynetwork.com/templates/Download. aspx?ContentItemId=891. P lease n o te that access to this content requires initial registration.

T h is m u ltim ed ia e x cu rs io n p ro v id es a n e x a m p le o f h o w a d d itional in form ation m ad e a v ailab le th ro u g h a n e n te rp rise data w a re h o u s e c a n assist in d e cisio n m aking.

The DSS-BI Connection B y n o w , y o u sh ou ld b e a b le to s e e s o m e o f th e sim ilarities a n d d iffe re n ce s b e tw e e n D SS a n d B I. First, th e ir arch ite ctu res are v e ry sim ilar b e c a u s e B I e v o lv e d fro m D SS. H ow ever, B I im p lies th e u s e o f a data w a re h o u s e , w h e re a s D SS m ay o r m ay n o t h a v e s u ch a feature. B I is, th e re fo re , m o re ap p rop riate fo r larg e o rg a n iz a tio n s (b e c a u s e data w a re h o u s e s are e x p e n s iv e to b u ild a n d m ain tain ), b u t D SS ca n b e a p p ro p riate to a n y typ e o f organization.

S e c o n d , m o st D SS are co n stru cte d to d irectly su p p o rt s p e c ific d ecisio n m aking. B I sy stem s, in g e n era l, a re g e a re d to pro v id e a ccu ra te an d tim ely in form ation, an d th e y sup­ p o rt d e c is io n su p p o rt in directly. T h is situ ation is ch a n g in g , h o w e v e r, as m o re a n d m o re d e c is io n su p p o rt to o ls a re b e in g ad d ed to B I so ftw a re p a ck a g e s.

Chapter 1 • An O verview o f B u sin ess Intelligence, Analytics, and D ecisio n Support 4 9

Third , B I h a s an e x e c u tiv e a n d strategy o rie n tatio n , e sp e cia lly in its B P M a n d d ash­ b oard co m p o n e n ts . D SS, in con trast, is o rie n te d to w ard analysts.

Fou rth, m o st B I sy stem s a re co n stru cte d w ith co m m e rcia lly a v ailab le to o ls a n d c o m ­ p o n en ts th at a re fitted to th e n e e d s o f o rgan ization s. In b u ild in g D SS, th e in te rest m ay b e in co n stru ctin g so lu tio n s to very u n stru ctu red p ro b lem s. In s u c h situ ation s, m o re p ro ­ gram m ing (e .g ., u sin g to o ls s u ch as E x c e l) m ay b e n e e d e d to cu sto m ize th e solu tions.

Fifth, D SS m e th o d o lo g ie s an d e v e n s o m e to o ls w e re d e v e lo p e d m o stly in th e a c a ­ d em ic w o rld . B I m e th o d o lo g ie s an d to o ls w e re d e v e lo p e d m o stly b y so ftw are co m p a n ie s. (S e e Z am an, 2 0 0 5 , fo r in form ation o n h o w B I h a s e v o lv e d .)

Sixth, m a n y o f th e to o ls that B I u se s a re a lso co n sid e re d D SS to o ls. F o r e x a m p le , data m in in g a n d p red ictiv e analysis are c o r e to o ls in b o th areas.

A lthough s o m e p e o p le e q u a te D SS w ith B I, th e s e sy stem s a re not, a t p re sen t, th e sam e. It is in te restin g to n o te that s o m e p e o p le b e lie v e that D SS is a p art o f B I — o n e o f its analytical to o ls . O th ers th in k that B I is a s p e c ia l c a s e o f D SS th at d eals m o stly w ith rep o rt­ ing, co m m u n ica tio n , an d co lla b o ra tio n (a fo rm o f d ata-o rien ted D SS). A n o th e r e x p la n a ­ tio n (W atso n , 2 0 0 5 ) is that B I is a result o f a co n tin u o u s re v o lu tio n and , a s s u ch , D SS is o n e o f B I ’s original e le m e n ts. In this b o o k , w e sep arate D SS fro m B I. H o w e v e r, w e p o in t to th e D S S -B I c o n n e c tio n freq u en tly . Further, a s n o te d in th e n e x t s e c tio n onw ard , in m any c irc lc s B I h a s b e e n su b su m e d b y th e n e w term a n a ly tics o r d a ta scien ce.

SECTION 1 .7 REVIEW QUESTIONS

1 . D e fin e BI. 2 . List an d d e s c r ib e th e m ajo r c o m p o n e n ts o f B I. 3. W h at a re th e m a jo r sim ilarities a n d d iffe ren ce s o f D SS a n d BI?

1,8 B U SIN ESS A N A L Y T IC S OVERVIEW T h e w o rd “an alytics” h a s re p la ced th e previou s individual co m p o n e n ts o f com p u terized d ecisio n su p p o rt te ch n o lo g ie s that h av e b e e n available u n d er v arious lab e ls in th e past. In d eed , m an y p ractitioners and a cad em ics n o w u se th e w o rd a n a ly tics in p la ce o f B I. Although m an y au th ors an d con su ltan ts h av e d efin ed it slightly differently, o n e ca n v ie w analytics as th e p ro cess o f d ev elo p in g a ctio n a b le d ecisio n s o r re co m m en d a tio n fo r action s b a se d u p o n insights g e n era ted fro m h istorical data. T h e Institute fo r O p e ra tio n s R esearch and M an ag em en t S c ie n c e (IN FO RM S) has cre a te d a m a jo r initiative to org an ize and p ro ­ m o te analytics. A cco rd in g to INFORMS, analytics re p re sen ts th e co m b in a tio n o f co m p u te r te ch n o lo g y , m a n a g e m e n t s c ie n c e te ch n iq u es, and statistics to solve real p ro b lem s. O f co u rse , m an y o th e r organ ization s h av e p ro p o se d th eir o w n interpretations a n d m otivation fo r analytics. F o r e x a m p le , SAS Institute In c. p ro p o s e d eig h t lev els o f an alytics th at b eg in w ith stand ard ized reports fro m a co m p u te r system . T h e s e rep o rts essen tially p rovide a s e n se o f w h a t is h a p p e n in g w ith an organization. A dditional te c h n o lo g ie s h av e e n a b le d us to cre a te m o re cu sto m ized reports that c a n b e g e n era ted o n a n ad h o c b asis. T h e n e x t e x te n sio n o f rep ortin g tak es us to o n lin e analytical p ro cessin g (O L A P )-ty p e q u eries that allow a u se r t o dig d e e p e r an d d eterm in e th e sp e cific so u rce o f c o n c e rn o r o p p o rtu n i­ ties. T e c h n o lo g ie s available to d ay c a n also autom atically issu e alerts fo r a d e c is io n m aker w h e n p e rfo rm a n ce issu es w arrant s u ch alerts. At a co n su m e r lev el w e s e e s u ch alerts fo r w e ath e r o r o th e r issues. B u t sim ilar alerts c a n a lso b e g e n era ted in s p e cific settin gs w h e n sales fall a b o v e o r b e lo w a certain lev el w ith in a certain tim e p eriod o r w h e n th e in v en toiy fo r a s p e cific p ro d u ct is running low . All o f th e se ap p licatio n s are m ad e p o ss ib le through analysis a n d q u e rie s o n data b e in g c o lle cte d b y a n organization. T h e n e x t le v e l o f analysis m ight entail statistical analysis to b e tte r un derstand pattern s. T h e s e c a n th e n b e ta k e n a step fu rther to d e v e lo p fo reca sts o r m o d e ls fo r p red ictin g h o w cu stom ers m ig h t re sp o n d to

5 0 Part I * D ecisio n M aking and Analytics: An O verview

FIGURE 1.5

L T a t o o d â w mSf ° r ° n80ing Service/Product « « * * * * ■ an organization o th e r r !c h ^ ! ! h,ap p em n « an d w h at is lUcely » h ap p e n , it c a n also em p loy

e r te ch n iq u es to m a k e th e b e s t d ecisio n s u n d e r th e circu m stan ces. T h e s e eigh t lev els o f

:2Z“LZ'X" “ - • * ' T h is id ea o f lo o k in g at all th e data to u n d erstan d w h a t is h a p p e n in g w h at will

h a p p e n , a n d h o w to m a k e th e b e s t o f it h a s a ls o b e e n e n ca p s u la te d b y IN FORM S in p ro p o sin g th ree lev els o f analytics. T h e s e th re e lev els a r e identified (in fo rm s.o r^ Community/Analytics) as d escrip tiv e, p red ictiv e, an d p re scrip tiv e . Fig u re 1.5 p r e s e n t

o grap h ical v iew s o f th e se th ree lev els o f analytics. O n e v ie w su gg ests that th e s e th ree

t T a ” T h e lderPendC nt ST ( 1 3 Iad der) ^ ° n e ^ ° f:analytlCS a p p lica tio n leads an o th er. T h e in te rco n n e cte d circle s v ie w su ggests th at th e re is actu ally s o m e ov erlap

* * * ? ° f analytlCS, In CaSe’ th e in te rc o n n e c te d n atu re o f d ifferent ty p e s o f an aly tics a p p lica tio n s is evident. W e n e x t in tro d u ce th e s e th re e lev els o f analytics.

D escriptive A nalytics

D escriptive or rep o rtin g an aly tics re fe rs to k n o w in g w h a t is h a p p e n in g in th e o r g a n iz a tio n a n d u n d e rs ta n d in g s o m e u n d e rly in g tre n d s a n d c a u s e s o f s u c h o c c u r ­ r e n c e s . T h is in v o lv e s , first o f all, c o n s o lid a tio n o f d ata s o u r c e s a n d a v ailab ility o f

P re d ictiv e Statistical Analysis and

Data Mining

R e p o rtin g Visualization

Periodic, ad hoc Reporting Trend Analysis

P r e s c r ip tiv e Management Science Models and Solution

Management Science Models and

Solution P red ictiu e

Statistical Analysis and

Data Mining R e p o rtin g

Visualization Periodic,

ad hoc Reporting Trend Analysis

Three Types of Analytics.

Chapter 1 • An O verview o f B usiness Intelligence, Analytics, and D ecisio n Support 51

a ll re le v a n t d a ta in a fo rm th at e n a b le s a p p r o p ria te re p o r tin g a n d a n a ly sis. U su ally d ev e lo p m e n t o f this d ata in fra stru ctu re is p a rt o f d ata w a r e h o u s e s , w h ic h w e stu d y in C h apter 3 . F ro m th is d ata in fra stru ctu re w e c a n d e v e lo p a p p r o p ria te re p o r ts , q u e rie s , -i.erts, a n d tre n d s u s in g v a rio u s re p o r tin g to o ls a n d te c h n iq u e s . W e s tu d y th e s e in C h ap ter 4.

A sig n ifican t te c h n o lo g y that h a s b e c o m e a k e y p la y er in this a re a is visualization . Using th e la test visu alization to o ls in th e m a rk e tp la ce , w e c a n n o w d e v e lo p p o w erfu l insights in to th e o p era tio n s o f o u r org anization. A p p licatio n C ase s 1 .2 an d 1 .3 highlight fo m e s u ch a p p lica tio n s in th e h e a lth ca re d om ain. C o lo r ren d erin g s o f s u ch ap p licatio n s are av ailab le o n th e c o m p a n io n W e b site an d a lso o n T a b le a u ’s W e b site. C h ap ter 4 co v ers v isu alization in m o re detail.

Application Case 1.2 Eliminating Inefficiencies at Seattle Children's Hospital Seattle C h ild ren ’s w as th e sev en th h ig h est ranked ch ild ren ’s h o sp ital in 2 0 1 1 , a cco rd in g to U.S. News & W orld R eport. F o r a n y org an izatio n th at is c o m ­ m itted to sav in g liv es, id entifying an d rem ov in g the in efficie n cie s fro m sy stem s and p ro c e s s e s s o that m o re re s o u rc e s b e c o m e a v ailab le to c a te r to p atien t care b e c o m e v e ry im portant. At Seattle C h ild ren’s, m an ag e m e n t is co n tin u o u sly lo o k in g fo r n e w w ay s to im p ro ve th e quality, safety , an d p ro c e s s e s from th e tim e a p a tie n t is ad m itted to th e tim e th e y are d ischarged . T o this end , th e y sp e n d a lo t o f tim e in analyzing th e d ata a s s o cia te d w ith th e p atie n t visits.

T o q u ick ly a i m p atien t and hospital data into insights, Seattle C hildren’s im plem en ted T a b le a u Softw are’s b u sin e ss in te llig e n ce ap p licatio n . It p ro­ vides a b ro w se r b a se d o n e asy -to -u se analytics to th e stakehold ers; this m ak e s it intuitive fo r individuals to create visualization s and to u n d erstan d w h at th e data has to offer. T h e data analysts, b u sin ess m anagers, an d financial analysts as w e ll a s clinician s, d octo rs, an d re sea rch ers a re all u sin g d escriptive analytics to solve d ifferen t p ro b lem s in a m u ch faster w ay. T h ey are d ev elo p in g visual system s o n th e ir ow n, resulting in d ash b o ard s an d sco re card s th at help in d efin in g th e standards, th e cu rren t p erfo rm an ce ach iev ed m e a su re d against th e standards, an d h o w th ese sy stem s w ill g ro w into the future. T h ro u g h th e use o f m o n th ly a n d daily d ashbo ard s, day-to-day d ecisio n m ak in g at Seattle C hildren’s has im proved significantly.

Seattle C h ild ren ’s m e a su re s p atien t w ait-tim es an d an aly zes th e m w ith th e h e lp o f visualization s to d isco v e r th e ro o t ca u se s an d con trib u tin g facto rs

fo r p atie n t w aiting. T h e y fo u n d that early d elays c a s c a d e d during th e day. T h e y fo c u s e d o n o n -tim e ap p o in tm en ts o f p a tie n t s e rv ice s as o n e o f th e s o lu ­ tio n s to im p ro vin g p atie n t ov erall w aitin g tim e an d in cre asin g th e availability o f b ed s. Seattle C hildren's s a v e d a b o u t $3 m illion fro m th e su p p ly ch ain , and w ith th e h e lp o f to o ls lik e T a b le a u , th e y a re find ­ ing n e w w ays to in c re a s e savings w h ile treatin g as m an y p a tie n ts as p o ss ib le b y m ak in g th e existin g p ro c e s s e s m o re efficien t.

Q u e s t i o n s f o r D i s c u s s i o n

1. W h o a re th e u se rs o f th e tool? 2. W h a t is a d ashboard? 3. H o w d o e s v isu alization h e lp in d e c is io n m aking? 4 . W h at a re th e sig n ifican t results a ch ie v e d b y th e

u se o f T ab leau ?

W h a t W e C a n L e a r n f r o m T h is A p p lic a tio n C a se

T h is A p p licatio n C a se s h o w s th at re p o rtin g analyt­ ics in volvin g v isu alization s s u c h as d ash b o ard s can o ffe r m a jo r insights in to e x istin g data an d s h o w h o w a variety o f u sers in d iffe ren t d om ain s an d d ep art­ m e n ts c a n co n trib u te to w ard p ro c e s s and q u a l­ ity im p ro v em en ts in a n organization. Fu rth erm ore, e x p lo rin g th e d ata visually c a n h e lp in identifying th e ro o t c a u se s o f p ro b le m s an d pro v id e a b a sis fo r w o rk in g to w ard p o s s ib le solu tions.

Source: Tableausoftware.com, “Eliminating Waste at Seattle Children’s, ” tableausoftware.com/ eliminating-waste-at-seattle- childrens (accessed Febmaiy 2013).

5 2 Part I • D ec isio n M aking and Analytics: An Overview

Application Case 1.3 Analysis at the Speed of Thought K aleid a H ealth , th e largest h e a lth ca re p ro v id e r in w e ste rn N ew Y o rk , h a s m o re th a n 1 0 ,0 0 0 e m p lo y ­ e e s , fiv e h o sp itals, a n u m b e r o f clin ics an d nu rsing h o m es, a n d a v isitin g-n u rse a sso cia tio n th at d eals w ith m illion s o f p atie n t reco rd s. K a leid a ’s traditional rep ortin g to o ls w e re in a d e q u a te to h a n d le th e g ro w ­ ing data, an d th e y w e re fa c e d w ith the c h a lle n g e o f fin d in g a b u sin e ss in te llig e n ce to o l th a t c o u ld h an d le large d ata s e ts effo rtlessly, q u ick ly , a n d w ith a m u ch d e e p e r an aly tic capability.

At K aleid a, m an y o f th e calcu latio n s are n o w d o n e in T a b le a u , prim arily pu lling th e d ata from O ra cle d a ta b a se s in to E x c e l an d im p orting th e d ata into T a b le a u . F o r m an y o f th e m o n th ly a n a ­ ly tic re p o rts, data is d irectly e x tra cted in to T ab leau fro m th e d ata w a re h o u s e ; m an y o f th e data q u eries a re sav e d a n d reru n, resultin g in tim e savings w h e n d ealin g w ith m illions o f re co rd s— e a c h h av in g m o re th an 4 0 field s p e r reco rd . B e s id e s sp e e d , K aleid a a lso u s e s T a b le a u to m e rg e d ifferen t ta b le s fo r g e n ­ e ratin g extracts.

U sing T a b le a u , K aleid a c a n an alyze e m e rg en c y ro o m data t o d eterm in e th e n u m b e r o f p atien ts w h o visit m o re th a n 10 tim es a year. T h e data o fte n reveal that p e o p le fre q u e n tly u se e m e rg e n c y ro o m and a m b u la n ce s e rv ice s in ap p ro p riately fo r sto m a ch ­ a c h e s , h e a d a c h e s , and fevers. K aleid a c a n m an ag e re s o u rce utilization s— th e u se a n d c o s t o f su p p lies— w h ich will u ltim ately le a d t o e fficie n cy a n d stand ard ­ izatio n o f su p p lie s m a n a g e m e n t a cro ss th e system .

K a leid a n o w h a s its o w n b u sin ess in te llig e n ce d ep artm en t an d u se s T a b le a u to co m p a re its e lf to

o th e r h osp itals a c ro ss th e cou ntry. C o m p ariso n s are m ad e o n v ario u s asp e cts, s u ch as len g th o f p atien t stay, h o sp ital p ra ctice s, m ark e t sh are, an d p artn er­ sh ip s w ith d o cto rs.

Q u e s t i o n s f o r D i s c u s s i o n

1. W h a t are th e d esire d fu n ctio n alitie s o f a rep o rt­ ing tool?

2. W h at ad v an tag e s w e re d eriv ed b y u sin g a rep o rt­ ing to o l in th e case?

W h a t W e C a n L e a r n f r o m T h is A p p lic a tio n C a se

C o rre ct s e le c tio n o f a rep ortin g to o l is e x trem ely im portant, e sp e c ia lly if a n org an izatio n w an ts to d eriv e v alu e fro m reporting. T h e g e n e ra te d reports an d visu a lizatio n s sh o u ld b e e asily d isce rn ib le; they sh o u ld h e lp p e o p le in d ifferen t s e c to rs m a k e s e n se o u t o f th e re p o rts, identify th e p ro b lem a tic areas, an d co n trib u te to w ard im p ro vin g them . M any future org an izatio n s w ill req u ire re p o rtin g an alytic to o ls that are fast an d c a p a b le o f h an d lin g h u ge am o u nts o f d ata e fficie n tly to g e n e ra te d esired rep o rts w ith­ o u t th e n e e d fo r third-party co n su ltan ts an d serv ice providers. A tru ly u se fu l re p o rtin g to o l c a n e x e m p t org an izatio n s fro m u n n e ce s s a ry exp en d itu re.

Source: TabIeau.software.com, “Kaleida Health Finds Efficiencies, Stays Competitive,” tableausoftware.com/leam/stories/user- experience-speed-thought-kaleida-health (accessed February 2013).

Predictive A nalytics Predictive analytics aim s to d eterm in e w h at is lik e ly to h a p p e n in th e future. T h is analy­ sis is b a s e d o n statistical te c h n iq u e s as w e ll a s o th e r m o re re ce n tly d e v e lo p e d te ch n iq u e s th at fall u n d e r th e g e n eral ca te g o ry o f d ata m ining . T h e g o al o f t h e s e te ch n iq u e s is to b e a b le to p re d ict i f th e cu sto m e r is lik ely to sw itch t o a co m p e tito r ( “ch u rn ”), w h a t th e cu s­ to m e r is lik e ly to b u y n e x t a n d h o w m u ch , w h at p ro m o tio n a cu sto m e r w o u ld re sp o n d to, o r w h e th e r this cu sto m e r is a cred itw orth y risk. A n u m b e r o f te ch n iq u e s a re u s e d in d e v e lo p in g p red ictiv e an aly tical ap p licatio n s, in clu d in g vario u s classificatio n algorithm s. F o r e x a m p le , as d e s crib e d in C h ap ters 5 an d 6 , w e c a n u s e classificatio n te c h n iq u e s su ch a s d e cisio n tre e m o d els a n d n eu ral n e tw o rk s to p re d ict h o w w e ll a m o tio n p ictu re will d o a t th e b o x o ffice . W e c a n a ls o u se clu sterin g algorithm s fo r s eg m en tin g cu sto m ers in to d ifferen t clu sters to b e a b le to targ et s p e c ific p ro m o tio n s to them . Finally, w e c a n

C hapter 1 • An O verview o f B u sin ess Intelligen ce, Analytics, and D ecisio n Support 5 3

use a s s o cia tio n m in in g te c h n iq u e s to e stim ate relatio n sh ip s b e tw e e n d iffe ren t p u rch asin g b ehav io rs. T h a t is, i f a cu sto m e r b u y s o n e p ro d u ct, w h a t e ls e is th e cu sto m e r lik e ly to pu r­ chase? S u c h a n aly sis c a n assist a retailer in re co m m e n d in g o r p ro m o tin g re la ted produ cts. F o r e x a m p le , an y p ro d u ct s e a rch o n A m a z o n .co m results in the re taile r a lso suggestin g other sim ilar p ro d u cts th at m ay in te rest a cu sto m er. W e will study th e s e te c h n iq u e s and th eir a p p lica tio n s in C h ap ters 6 throu gh 9- A p p lication C ases 1.4 an d 1 .5 hig h lig h t so m e sim ilar a p p licatio n s. A p p lication C ase 1 .4 in tro d u ces a m o v ie y o u m ay h av e h e ard of: M oneyball. It is p e rh a p s o n e o f th e b e s t e x a m p le s o f ap p licatio n s o f p re d ictiv e analysis in sports.

Application Case 1.4 M oneyball: Analytics in Sports and Movies M oneyball, a b io g ra p h ica l, sp orts, dram a film , w as re le a se d in 2011 and d irecte d b y B e n n e tt M iller. T h e film w a s b a s e d o n M ich ael Lew is’s b o o k , M oneyball. T h e m o v ie g a v e a d eta iled a c c o u n t o f th e O ak lan d A thletics b a s e b a ll te a m during th e 2 0 0 2 s e a s o n and the O a k la n d g e n eral m an ag e r’s e ffo rts to a sse m b le a com p etitiv e team .

T h e O ak lan d A thletics su ffered a b ig lo ss to th e Newr Y o rk : Y a n k e e s in 20 0 1 p o stse a so n . As a result, O ak la n d lo s t m a n y o f its star p layers to fre e a g e n c y and e n d e d u p w ith a w e a k te a m w ith u n fav o rab le financial p ro sp ects. T h e g e n eral m an ag e r’s efforts to re a sse m b le a co m p etitiv e te a m w e r e d en ie d b e c a u s e O ak lan d h a d lim ited payroll. T h e sco u ts for th e O ak lan d A thletics fo llo w e d th e o ld b a se b a ll cu sto m o f m ak in g s u b je ctiv e d e cis io n s w h en s e le ctin g th e team m e m b e rs. T h e g e n e ra l m an ag e r th e n m e t a you ng, co m p u te r w h iz w ith an e c o n o m ic s d eg ree from Y a le . T h e g e n e ra l m an ag e r d e c id e d to ap p o in t him as th e n e w assistan t g e n e ra l m an ager.

T h e assistan t g e n eral m an ag e r h a d a d e e p p as­ sio n fo r b a s e b a ll an d h ad th e e x p e rtis e to c m n c h the n u m b e rs fo r th e g am e. H is lo v e fo r th e g a m e m ad e him d e v e lo p a rad ical w a y o f un d erstand ing b a se b a ll statistics. H e w a s a d iscip le o f B ill J a m e s , a m arginal fig u re w h o o ffe red ratio n alized te ch n iq u e s io an aly ze b a se b a ll. Ja m e s lo o k e d at b a s e b a ll statis­ tics in a d iffe ren t w ay , cru n ch in g th e n u m b e rs pu rely o n facts a n d e lim in atin g subjectivity. Ja m e s p io ­ n e e re d th e n on trad ition al analysis m e th o d calle d th e S ab erm etric ap p ro a ch , w h ich d eriv ed fro m SABR— S o cie ty fo r A m erican B a s e b a ll R esearch .

T h e a s s is ta n t g e n e r a l m a n a g e r fo llo w e d th e S a b e rm e tric a p p r o a c h b y b u ild in g a p re d ic tio n

m o d e l to h e lp th e O a k la n d A th le tics s e l e c t p la y ­ e rs b a s e d o n th e ir “o n - b a s e p e r c e n ta g e ” ( O B P ) , a s ta tistic th a t m e a s u re d h o w o fte n a b a tte r r e a c h e d b a s e fo r a n y r e a s o n o t h e r th a n fie ld in g e rro r, fie ld ­ e r ’s c h o ic e , d ro p p e d / u n ca u g h t third s trik e , fie ld e r ’s o b s tru c tio n , o r c a tc h e r ’s in te r fe r e n c e . R a th e r th an re ly in g o n th e s c o u t ’s e x p e r ie n c e a n d in tu itio n , th e a s s is ta n t g e n e r a l m a n a g e r s e le c te d p la y e rs b a s e d a lm o s t e x c lu s iv e ly o n O B P .

S p o iler A lert: T h e n e w te a m b e a t all od d s, w o n 2 0 co n s e cu tiv e g a m e s , an d s e t a n A m erican L eagu e record .

Q u e s t i o n s f o r D i s c u s s i o n

1. H o w is p re d ictiv e an alytics a p p lie d in M oneyball? 2. W h at is th e d iffe re n ce b e tw e e n o b je c tiv e and

s u b je ctiv e a p p r o a c h e s in d e c is io n m aking?

W h a t W e C a n L e a m f r o m T h is A p p lic a tio n C a se

A n aly tics fin d s its u s e in a v arie ty o f ind u stries. It h e lp s o rg a n iz a tio n s re th in k th e ir trad itio n al p r o b ­ le m -s o lv in g a b ilitie s, w h ic h are m o st o f te n s u b je c ­ tiv e, re ly in g o n th e sa m e o ld p r o c e s s e s to find a so lu tio n . A n aly tics ta k e s th e rad ical a p p r o a c h o f u s in g h isto rica l d ata to fin d fa c t-b a s e d so lu tio n s th a t w ill re m a in a p p ro p ria te fo r m ak in g e v e n future d e c is io n s .

Source: Wikipedia, “On-Base Percentage,” en.wikipedia.org/ wiki/On_base_percentage (accessed January 2013); Wikipedia, “Saberrnetricsm,” wikipedia.org/wiki/Saberm etfics (accessed January 2013).

Part I • D e c is io n M aking and Analytics: An Overview

Application Case 1.5 Analyzing Athletic Injuries A ny a th le tic activ ity is p r o n e to in ju rie s. I f t h e in ju ­ ries a re n o t h a n d le d p ro p e rly , th e n th e te a m suf­ fers. U sin g a n a ly tic s to u n d e rstan d in ju rie s c a n h e lp in d eriv in g v a lu a b le in sig h ts th a t w o u ld e n a b le th e c o a c h e s a n d te a m d o c to rs to m a n a g e th e te a m c o m p o s itio n , u n d e rsta n d p la y e r p ro file s, a n d ulti­ m a te ly a id in b e tte r d e c is io n m ak in g c o n c e rn in g w h ic h p la y e r s m ig h t b e a v a ila b le to p la y at an y g iv e n tim e.

In a n e x p lo ra to ry study, O k la h o m a State U niversity a n a ly z ed A m erican fo o tb a ll-re la ted sport in ju ries b y u sin g re p o rtin g and p red ictiv e analytics. T h e p ro je c t fo llo w e d th e CRISP-D M m e th o d o l­ o g y to u n d erstan d th e p ro b le m o f m ak in g re c o m ­ m e n d atio n s o n m an ag in g inju ries, un derstand ing th e vario u s d ata e le m e n ts c o lle c te d a b o u t injuries, c le a n in g th e d ata, d ev elo p in g visu alization s to draw v arious in fe re n c e s , b u ild in g p red ictiv e m o d e ls to an aly ze th e injury- h e a lin g tim e p erio d , a n d draw ing s e q u e n c e ru le s to p re d ict th e re latio n sh ip am o n g th e in ju ries a n d th e v ario u s b o d y part parts afflicted w ith injuries.

T h e inju ry data s e t c o n s is te d o f m o re than 5 6 0 fo o tb a ll in ju ry re co rd s, w h ich w e re ca te g o riz e d in to in ju ry -sp e cific v ariab les— b o d y part/site/later- ality, a ctio n ta k e n , sev erity, injury ty p e, injury start an d h e a lin g d ates— and p layer/ sp ort-sp ecific varia­ b le s — p lay er ID , p o sitio n p lay ed , activity, o n s e t, an d g a m e lo c a tio n . H e alin g tim e w as ca lcu la ted fo r e a c h re co rd , w h ic h w a s cla ssifie d in to d ifferen t sets o f tim e p e rio d s: 0 - 1 m o n th , 1 - 2 m o n th s, 2 - 4 m onths, 4 - 6 m o n th s, a n d 6 - 2 4 m onths.

V ario u s visu alization s w e re built to draw in fe re n c e s fro m injury d ata s e t in form ation d ep ict­ ing th e h e a lin g tim e p e rio d a s s o cia te d w ith p lay ers’ p o sitio n s, sev erity o f inju ries a n d th e h e a lin g tim e p e rio d , tre a tm e n t o ffe re d an d th e a s s o cia te d h e a lin g tim e p erio d , m a jo r inju ries afflictin g b o d y parts, an d s o forth.

N eural n e tw o rk m o d els w e re b u ilt to p re ­ d ict e a c h o f th e h e a lin g ca te g o rie s using IB M SPSS M o d eler. S o m e o f th e p re d icto r v ariab les w e re cu r­ rent statu s o f injury, severity, b o d y part, b o d y site, ty p e o f injury, activity, e v e n t lo ca tio n , a c tio n tak e n , a n d p o sitio n p lay ed . T h e s u c c e s s o f classify in g th e h e a lin g ca te g o ry w a s q u ite g o o d : A ccu racy w a s 79-6 p e rce n t. B a s e d o n th e analy sis, m an y b u sin e ss re c ­ o m m en d a tio n s w e re su g g e ste d , in clu d in g e m p lo y ­ ing m o re sp e cia lists’ input fro m in ju iy o n s e t instead o f lettin g th e training ro o m sta ff s c r e e n th e in ju red p layers; training p lay ers a t d efen siv e p o sitio n s to a v o id b e in g in ju red ; a n d h o ld in g p ractice to th o r­ ou g h ly s a fe ty -c h e c k m ech an ism s.

Q u e s t i o n s f o r D i s c u s s i o n

1. W hat ty p e s o f an aly tics are ap p lie d in th e injury analysis?

2. H o w d o v isu alization s aid in u n d erstan d in g th e d ata and d eliv erin g insights in to th e data?

3. W h at is a cla ssifica tio n problem ? 4. W h at c a n b e d eriv ed b y p erfo rm in g s e q u e n c e

analysis?

W h a t W e C an L e a r n f r o m T h is A p p lic a tio n C a se

F o r an y an alytics p ro je ct, it is alw ays im portant to u n d erstan d th e b u sin e ss d o m ain a n d th e cu r­ re n t state o f th e b u sin e ss p ro b le m th ro u g h e x te n ­ siv e analysis o f th e only re so u rce — h istorical data. V isu alizatio n s o fte n pro v id e a g reat to o l fo r g aining th e initial insights in to d ata, w h ich c a n b e further refin ed b a s e d o n e x p e r t o p in io n s to id en tify th e re la ­ tive im p o rtan ce o f th e d ata e le m e n ts related to th e p ro b lem . V isu alizatio n s a lso aid in g e n eratin g ideas fo r o b s c u re b u s in e s s p ro b le m s, w h ic h c a n b e pu r­ su e d in b u ild in g p red ictiv e m o d els th at co u ld h e lp o rg an ization s in d e c is io n m aking.

Prescriptive A nalytics T h e third ca te g o ry o f a n alytics is term ed p rescrip tive analytics. T h e g o a l o f p rescriptive an aly tics is to re co g n iz e w h a t is g o in g o n as w e ll as th e lik e ly fo re c a s t a n d m ak e d ecisio n s to a c h ie v e th e b e s t p e rfo rm a n ce p o ssib le . T h is g ro u p o f te c h n iq u e s h a s historically b e e n stu d ied u n d er th e u m b rella o f o p era tio n s re se a rch o r m an a g e m e n t s c ie n c e s an d h a s g e n ­ erally b e e n aim ed a t op tim izing th e p e rfo rm a n ce o f a system . T h e g o a l h e re is to p rovide

Chapter 1 • An O verview o f B u sin ess Intelligence, Analytics, and D ecision Support 55

2 decision o r a re co m m e n d a tio n fo r a s p e cific a ctio n . T h e s e re co m m en d a tio n s c a n b e in i h e form s o f a s p e c ific yes/no d e cisio n fo r a p ro b le m , a s p e c ific am o u n t (say , p ric e fo r a so e cific item o r airfare to c h a rg e ), o r a c o m p le te s e t o f p ro d u ctio n plans. T h e d ecisio n s m ay b e p re se n te d t o a d e c is io n m a k e r in a report o r m ay d irectly b e u s e d in a n au tom ated decision ru les sy stem (e .g ., in airline p ricin g sy stem s). T h u s, th e s e ty p e s o f an aly tics ca n u o b e te rm e d d ecision o r norm ative analytics. A p p lication C ase 1.6 gives a n e x a m p le r f s u ch p rescrip tiv e an alytic a p p licatio n s. W e w ill le a rn a b o u t s o m e o f th e s e te c h n iq u e s irid sev eral ad d ition al a p p lica tio n s in C h ap ters 10 th ro u g h 12.

Application Case 1.6 Industrial and Commercial Bank of China (ICBC) Employs Models to Reconfigure Its Branch Network T h e Industrial an d C o m m ercial B a n k o f China (IC B C ) has m o re th an 1.6,000 b ra n c h e s and serves o v er 2 3 0 m illion individual cu stom ers an d 3 .6 mil­ lion co rp o ra te clien ts. Its daily financial tran saction s total a b o u t $ 1 8 0 m illion. It is also th e largest p u b ­ licly trad ed b a n k in th e w o rld in term s o f m arket capitalization, d ep o sit v o lu m e, and profitability. T o stay com p etitiv e an d in cre a se profitability, IC B C w as faced w ith th e c h a lle n g e to q u ick ly ad ap t to th e fast- p a ce d e c o n o m ic grow th, u rbanization, an d in crease in p ersonal w e a lth o f the C h in ese. C h an ges h a d to b e im plem ented in o v e r 3 0 0 cities w ith high variability in cu stom er b e h a v io r an d financial status. O bv io u sly, th e natu re o f th e ch a lle n g e s in s u ch a h u g e e c o n o m y m ean t th at a larg e-scale op tim ization so lu tio n h ad to b e d e v e lo p e d to lo ca te b ra n c h e s in th e right places, with right serv ices, to serve th e right cu stom ers.

W ith their existing m ethod , ICBC u sed to d ecid e w h ere to o p e n n e w b ran ch es through a scoring m odel in w hich d ifferent variables w ith varying w eight w ere u sed as inputs. S o m e o f th e variables w e re cu stom er flow , n u m ber o f residential hou seh old s, and nu m ber o f com petitors in th e in ten d ed geograp h ic region. This m ethod w a s d eficien t in determ ining th e cu stom er dis­ tribution o f a g eo g rap h ic area. T h e existing m ethod w as also u n a b le to optim ize th e distribution o f b an k bran ch es in th e b ran ch netw ork. W ith supp ort from IBM, a b ran ch reconfigu ration (B R ) to o l w as devel­ op ed . Inputs fo r th e B R system are in th ree parts:

a. G e o g ra p h ic data w ith 8 3 d ifferen t categ o ries b. D e m o g ra p h ic and e c o n o m ic d ata w ith 22 dif­

fe re n t ca te g o rie s c . B ra n c h tran sactio n s a n d p e rfo rm a n ce d ata that

co n s is te d o f m o re th a n 6 0 m illio n tran sactio n re co rd s e a c h day

T h e s e th ree inputs h e lp e d g e n era te a ccu rate cu s­ to m er d istribution fo r e a c h a re a and, h e n c e , h e lp e d th e b a n k o p tim ize its b ra n ch n etw o rk . T h e B R sy stem co n siste d o f a m ark e t p o ten tial calcu latio n m o d el, a b ra n ch n e tw o rk o p tim ization m o d el, a n d a b ra n ch site evalu ation m o d e l. In th e m ark e t p o ten tial m odel, th e cu sto m e r v o lu m e an d v a lu e is m ea su red b a se d o n input data a n d e x p e rt k n o w le d g e. F o r in stan ce, e x p e rt k n o w le d g e w o u ld h e lp d eterm in e if p e r­ so n al in co m e sh ou ld b e w e ig h te d m o re th an g ro ss d o m e stic p ro d u ct (G D P ). T h e g e o g ra p h ic are a s are also d em arcated into cells, and th e p re fe re n c e o f o n e ce ll o v e r th e o th e r is d eterm ined . In th e b ra n ch n e t­ w o rk op tim ization m o d el, m ix e d in te g er p rogram ­ m ing is u sed to lo c a te b ra n ch e s in can d id ate cells so that th e y co v e r th e largest m arket p o ten tial areas. In th e b ra n ch site e v alu atio n m o d el, th e value fo r estab lish in g b a n k b ra n c h e s at sp e cific lo ca tio n s is d eterm ined .

S in c e 2 0 0 6 , th e d e v e lo p m e n t o f th e B R has b e e n im p ro v e d th r o u g h a n itera tiv e p r o c e s s . IC B C ’s b r a n c h re c o n fig u r a tio n to o l h a s in c r e a s e d d e p o s its b y $ 2 1 .2 b illio n s in c e its in c e p tio n . T h is in c re a s e in d e p o s it is b e c a u s e th e b a n k c a n n o w re a c h m o re c u s to m e rs w ith th e rig h t s e r v ic e s b y u s e o f its o p tim iz a tio n to o l. In a s p e c ific e x a m p le , w h e n B R w a s im p le m e n te d in S u z h o u in 2 0 1 0 , d e p o s its in c re a s e d to $ 1 3 -6 7 b illio n fro m an in itial le v e l o f $ 7 .5 6 b illio n in 2 0 0 7 . H e n c e , th e B R to o l a s s is te d in a n in c r e a s e o f d e p o s its to th e tu n e o f $ 6 .1 1 b illio n b e tw e e n 2 0 0 7 a n d 2 0 1 0 . T h is p r o je c t w a s s e le c te d a s a fin a lis t in th e E d e lm a n C o m p e titio n 2 0 1 1 , w h ic h is ru n b y IN FO R M S t o p r o m o te actu al a p p lic a tio n s o f m a n a g e m e n t s c ie n c e / o p e ra tio n s re s e a r c h m o d e ls.

0C on tin u ed )

Part I • D ec isio n M aking and Analytics: An O verview

Application Case 1.6 (Continued) Q u e s t i o n s f o r D i s c u s s i o n

1. H o w c a n an aly tical te c h n iq u e s h e lp o rg an iza­ tio n s to re tain co m p etitiv e advantage?

2. H o w c a n d escrip tiv e a n d p red ictiv e analytics h e lp in p u rsu in g p rescrip tiv e analytics?

3. W h a t k in d s o f p rescrip tive an aly tic te ch n iq u e s a re e m p lo y e d in th e c a s e saidy?

4. A re th e p re scrip tiv e m o d e ls o n c e built g o o d forever?

W h a t W e C a n L e a r n f r o m T h is A p p lic a tio n C a se

M any o r g a n iz a tio n s in th e w o rld a re n o w e m b r a c ­ in g a n a ly tic a l te c h n iq u e s to s ta y c o m p e titiv e a n d a c h ie v e g ro w th . M any o rg a n iz a tio n s p ro v id e

c o n s u ltin g s o lu tio n s to th e b u s in e s s e s in e m p lo y ­ in g p re s c rip tiv e a n a ly tica l s o lu tio n s . It is e q u a lly im p o rta n t to h a v e p ro a c tiv e d e c is io n m a k e rs in th e o rg a n iz a tio n s w h o a re a w a re o f th e c h a n g in g e c o ­ n o m ic e n v ir o n m e n t as w e ll as th e a d v a n c e m e n ts in th e fie ld o f a n a ly tic s t o e n s u re th a t a p p ro p ria te m o d e ls a re e m p lo y e d . T h is c a s e s h o w s a n e x a m p le o f g e o g r a p h ic m a rk e t s e g m e n ta tio n a n d c u s to m e r b e h a v io ra l s e g m e n ta tio n te c h n iq u e s to is o la te th e p ro fita b ility o f c u s to m e rs a n d e m p lo y o p tim iz a tio n te c h n iq u e s to lo c a t e th e b r a n c h e s th a t d e liv e r high p ro fita b ility in e a c h g e o g r a p h ic s e g m e n t.

Source: X. Wang et al., “Branch Reconfiguration Practice Through Operations Research in Industrial and Commercial Bank of China,” Interfaces, January/February 2012, Vol. 42, No. 1, pp. 33-44: DOI: 10.1287/inte.1110.0614.

A nalytics Applied to D ifferent Domains A p p lication s o f an alytics in vario u s industry se cto rs h a v e sp a w n ed m an y re la ted a re a s o r at le a s t b uzzw ord s. It is alm o st fa s h io n a b le to a tta ch th e w o rd an a ly tics to any sp e cific indu stry o r typ e o f data. B e s id e s th e g e n e ra l c a te g o ry o f te x t analytics— aim ed at getting v alu e o u t o f te x t (to b e stu d ied in C h ap ter 6 )— o r W e b analytics— analyzing W e b data stream s (C h ap ter 7 )— m an y industry- o r p ro b le m -s p e c ific an alytics professions/stream s have c o m e u p . E x a m p le s o f s u ch are a s are m arketin g analytics, retail analytics, frau d ana­ lytics, tran sp ortation analytics, h e a lth analytics, sp o rts analytics, talen t analytics, b e h a v ­ ioral analytics, an d s o fo rth. F o r e x a m p le , A p p licatio n C ase 1.1 co u ld a lso b e term ed as a c a s e stud y in airlin e analytics. A p p lication C ases 1 .2 an d 1.3 w o u ld b e lo n g to h ealth analytics; A p p lication C ases 1 .4 an d 1.5 to sports an aly tics; A p p lication C ase 1 .6 to b an k analytics; a n d A p p lication C ase 1.7 to retail analy tics. T h e E n d -o f-C h ap ter A p p lication C ase co u ld b e term ed in su ra n ce analytics. Literally, a n y sy stem atic an alysis o f d ata in a s p e cific s e c to r is b e in g la b e le d as “(fill-in -b la n k s)” A nalytics. A lthough this m ay result in o v ersellin g th e c o n c e p ts o f a nalytics, th e b e n e fit is th a t m o re p e o p le in s p e c ific industries are aw are o f th e p o w e r and p o ten tial o f analytics. It a lso p ro v id e s a fo cu s to p ro fe ssio n als d ev elo p in g a n d ap p ly in g th e c o n c e p ts o f an alytics in a vertical s ecto r. A lthough m an y o f the te ch n iq u e s to d ev elo p a n alytics ap p lica tio n s m ay b e co m m o n , th e re a re u n iq u e issu es w ith in e a c h v e rtical s e g m e n t th at in flu e n ce h o w th e data m ay b e c o lle c te d , p ro cesse d , an aly zed , a n d th e ap p licatio n s im p lem en ted . T h u s, th e d ifferen tiatio n o f an alytics b a se d o n a vertical fo c u s is g o o d fo r th e ov erall gro w th o f th e d iscip line.

A nalytics or Data Science? E v e n as th e c o n c e p t o f analytics is g ettin g p o p u la r am o n g industry an d a c a d e m ic circles, an o th e r term h a s alread y b e e n in tro d u ced an d is b e c o m in g p o p u lar. T h e n e w term is d a ta scien ce. T h u s th e p ractition ers o f d ata s c ie n c e are d ata scien tists. Mr. D . J . Patil o f L in ked ln is s o m e tim e s cred ited w ith cre a tin g th e te rm d a ta sc ien c e. T h e r e h av e b e e n s o m e attem pts to d e s c rib e th e d iffe re n ce s b e tw e e n d ata analysts a n d data scien tists (e .g ., s e e this stud y at em c.com /collateral/ab ou t/new s/em c-d ata-scien ce-stu d y-w p.p d f). O n e v iew is that

Chapter 1 • An O verview o f B u sin ess Intelligence, Analytics, and D ecision Support 57

ia ia an aly st is ju s t a n o th e r term fo r p ro fe ssio n a ls w h o w e re d o in g b u sin e ss in te llig e n ce in iie form o f data co m p ila tio n , cle a n in g , rep ortin g, and p e rh a p s s o m e v isu alization . T h e ir s i l l sets in clu d e d E x c e l, s o m e SQ L k n o w le d g e , and rep ortin g. A re a d e r o f S e c tio n 1.8 srould re co g n iz e th a t a s d escrip tive o r rep ortin g analy tics. In con trast, a d ata s cie n tist is

resp o n sib le fo r p re d ictiv e analysis, statistical analy sis, and m o re a d v a n ce d an aly tical to o ls and algorithm s. T h e y m ay h av e a d e e p e r k n o w le d g e o f algorithm s and m ay re co g n iz e *b em u n d e r v ario u s la b e ls — d ata m ining, k n o w le d g e d isco v er)7, m a ch in e learn in g , and

forth. S o m e o f th e s e p ro fe ssio n a ls m ay also n e e d d e e p e r p ro g ram m in g k n o w le d g e to be a b le to w rite c o d e fo r d ata c le a n in g an d analysis in cu rre n t W e b -o rie n te d lan g u ag es 5-ich as Ja v a a n d P y th on . Again, o u r re a d ers sh o u ld re co g n iz e th e s e as fallin g u n d er the pred ictive a n d p rescrip tiv e an aly tics u m brella. O u r v ie w is that th e d istin ction b e tw e e n analytics an d d ata s c ie n c e is m o re o f a d e g re e o f te ch n ica l k n o w le d g e an d skill sets than d ie fu n ction s. It m ay a lso b e m o re o f a d istin ction a cro ss d iscip lin es. C o m p u te r s c ie n c e , statistics, a n d a p p lie d m ath em atics p rogram s a p p e a r to p re fe r th e data s c ie n c e lab el, reserving th e an aly tics la b e l fo r m o re b u s in ess-o rie n te d p ro fessio n als. As a n o th e r e x a m p le

this, ap p lie d p h y s ics p ro fe ssio n a ls h a v e p ro p o s e d u sin g n etw ork sc ien c e a s th e term for d escrib in g an alytics that relate to a g ro u p o f p e o p le — so cia l n e tw o rk s, su p p ly ch ain n etw o rks, an d s o forth. S e e barabasilab .n eu .ed u/n etw ork scien ceb ook /d ow n lPD F . fcdnl fo r a n ev o lv in g te x tb o o k o n this to p ic.

Aside fro m a clea r d ifferen ce in th e skill sets o f p ro fession als w h o o n ly h av e to d o iescriptive/reporting analytics versus th o se w h o e n g ag e in all three types o f analytics, the distinction is fu zzy b e tw e e n th e tw o lab els, at b est. W e o b serv e that grad uates o f ou r analytics program s ten d to b e re sp o n sib le fo r tasks m o re in line w ith data s c ie n c e p ro fes­ sionals (a s d efin ed b y so m e circles) than ju st reporting analytics. T h is b o o k is clearly aim ed i : introducing th e capabilities an d fu nctionality o f all analytics (w h ich in clu d es data sci­ e n ce), n o t ju st rep ortin g analytics. From n o w o n , w e will u se th ese term s interchangeably.

SECTION 1 .8 REVIEW QUESTIONS

1 . D efin e an aly tics. 2 . W hat is d escriptive analytics? W h at various to o ls are em p loyed in descriptive analytics?

3. H o w is d escrip tiv e an alytics d ifferen t fro m trad itional reporting? 4 . W h a t is a d a ta w a r e h o u s e ? H o w c a n d ata w a r e h o u s in g te c h n o lo g y h e lp in e n a ­

b lin g a n aly tics?

5. W h at is p re d ictiv e analytics? H o w c a n o rg an ization s e m p lo y p red ictiv e analytics? 6. W h a t is p re s c rip tiv e an aly tics? W h a t k in d s o f p r o b le m s c a n b e so lv e d b y p re s c rip ­

tive analytics?

7. D e fin e m o d e lin g fro m th e an alytics p e rsp ectiv e. 8 . Is it a g o o d id e a to fo llo w a h iera rch y o f d escrip tiv e an d p re d ictiv e an aly tics b e fo re

ap p lyin g p rescrip tiv e analytics? 9. H o w c a n an aly tics aid in o b je c tiv e d e c is io n m aking?

1.9 B R IE F IN TRO D U CTIO N TO B IG D A T A A N A LY T IC S W hat Is Big D ata? O u r b ra in s w o r k e x tr e m e ly q u ic k ly a n d a re e ffic ie n t a n d v e rsa tile in p r o c e s s in g la rg e am o u n ts o f all k in d s o f d ata: im a g es, te x t, so u n d s, sm e lls, a n d v id eo . W e p r o c e s s all d ifferen t fo rm s o f d a ta re la tiv ely easily . C o m p u ters, o n th e o th e r h a n d , a re still fin d in g it hard to k e e p u p w ith th e p a c e at w h ic h d ata is g e n e r a te d — le t a lo n e a n a ly z e it q u ick ly . W e h av e th e p r o b le m o f B ig D ata. S o w h a t is B ig D ata? S im p ly pu t, it is d ata th a t c a n n o t

M aking and Analytics: An Overview

b e s to r e d in a sin g le s to ra g e u n it. B ig D ata ty p ically re fe rs to d ata th a t is arrivin g in m a n y d iffe re n t fo rm s, b e th e y stru ctu red , u n stru ctu re d , o r in a stream . M ajor s o u rc e s o f s u ch d ata a re clic k s tre a m s fro m W e b sites, p o s tin g s o n s o c ia l m e d ia site s s u ch as F a c e b o o k , o r d ata fro m traffic, s e n s o rs , o r w e a th e r. A W e b s e a r c h e n g in e lik e G o o g le n e e d s to s e a r c h a n d in d e x b illio n s o f W e b p a g e s in o r d e r to g iv e y o u re le v a n t s e a r c h re su lts in a fra c tio n o f a s e c o n d . A lth o u g h this is n o t d o n e in re a l tim e , g e n e ra tin g an in d e x o f all th e W e b p a g e s o n th e In te rn e t is n o t a n e a s y task . L u ckily fo r G o o g le , it w a s a b le to s o lv e th is p ro b le m . A m o n g o th e r to o ls , it h a s e m p lo y e d B ig D ata a n aly tical te c h n iq u e s .

T h e r e are tw o a sp e cts to m an ag in g d ata o n this s c a le : sto rin g a n d p ro cessin g . I f w e co u ld p u rch a s e a n e x tre m e ly e x p e n s iv e s to ra g e s o lu tio n to sto re all th e data a t o n e p la ce o n o n e u n it, m ak in g this u n it fau lt to le ra n t w o u ld in v o lv e m a jo r e x p e n s e . An in g en iou s s o lu tio n w a s p ro p o s e d that involved sto rin g this data in ch u n k s o n d ifferen t m a ch in e s c o n n e c te d b y a n e tw o rk , pu tting a c o p y o r tw o o f th is ch u n k in d ifferen t lo ca tio n s o n th e n e tw o rk , b o th lo g ically an d p hy sically. It w as o rig in ally u se d at G o o g le (th e n calle d G oogle F ile System) and la te r d e v e lo p e d a n d re le a s e d as a n A p a ch e p ro je c t as th e H ad oo p D istributed File System (H D FS).

H o w ev er, sto rin g this d ata is o n ly h a lf the p ro b le m . D ata is w o rth le ss i f it d o e s n o t p ro v id e b u sin ess v alu e , an d fo r it to pro v id e b u s in e s s v alu e, it h a s to b e analyzed. H o w are s u ch vast am o u n ts o f data analyzed? P assin g all co m p u tatio n to o n e p o w erfu l co m p u te r d o e s n o t w o rk ; this s c a le w o u ld cre a te a h u g e o v e rh e a d o n s u ch a p o w e r­ fu l co m p u ter. A n o th er in g en io u s so lu tio n w a s p ro p o se d : P u sh co m p u ta tio n to th e data, in stead o f p u sh in g d ata to a co m p u tin g n o d e . T h is w a s a n e w parad igm , a n d it g a v e rise to a w h o le n e w w a y o f p ro c e s s in g d ata. T h is is w h at w e k n o w to d ay as th e M ap R ed u ce p ro g ram m in g parad igm , w h ich m ad e p ro ce ssin g B ig D a ta a reality. M ap R ed u ce w a s origi­ nally d e v e lo p e d at G o o g le , an d a s u b s e q u e n t v e rsio n w a s re le a s e d b y th e A p a ch e p ro je ct c a lle d H ad o o p M ap R ed uce.

T o d ay , w h e n w e talk a b o u t storing, p ro cessin g , o r analyzing B ig D ata, H D FS and M ap R ed u ce a re in v o lv ed at s o m e lev el. O th e r relev an t stand ard s an d so ftw are so lu tion s h av e b e e n p ro p o se d . A lthough th e m a jo r to o lk it is av a ila b le as o p e n so u rce , sev eral c o m p a n ie s h av e b e e n la u n ch e d to pro v id e train in g o r s p e cia liz e d analytical hard w are o r so ftw are s erv ices in this s p a c e . S o m e e x a m p le s are H orton W o rk s, C lou d era, a n d Terad ata Aster.

O v e r th e p a s t fe w y e a rs , w h a t w a s c a lle d B ig D a ta c h a n g e d m o re an d m o re as B ig D ata a p p lic a tio n s a p p e a re d . T h e n e e d to p r o c e s s data c o m in g in at a rap id rate ad d e d v e lo c ity to th e e q u a tio n . O n e e x a m p le o f fa st d ata p r o c e s s in g is a lg o rith m ic trad ing. It is th e u s e o f e le c tr o n ic p latfo rm s b a s e d o n a lg o rith m s fo r trad in g s h a re s o n th e fin an cial m ark et, w h ic h o p e ra te s in th e o r d e r o f m ic ro s e c o n d s . T h e n e e d to p r o c e s s d iffe re n t k in d s o f d ata a d d e d v arie ty to th e e q u a tio n . A n o th e r e x a m p le o f th e w id e v arie ty o f d ata is s e n tim e n t analy sis, w h ic h u s e s v ario u s fo rm s o f d a ta fro m s o c ia l m ed ia p latfo rm s a n d c u s to m e r re s p o n s e s to g a u g e s en tim en ts. T o d a y B ig D a ta is a s s o c ia te d w ith a lm o st a n y k in d o f la rg e d ata th a t h a s th e c h a ra c te ris tic s o f v o lu m e , v e lo city , a n d variety. A p p lica tio n C ase 1 .7 illu strates o n e e x a m p le o f B ig D a ta a n aly tics. W e w ill stu d y B ig D ata c h a ra c te ris tic s in m o re d etail in C h ap ters 3 a n d 13.

SECTION 1 .9 REVIEW QUESTIONS

1 . W h at is B ig D ata analytics?

2 . W h at are th e so u rce s o f B ig Data?

3 . W h at are th e ch aracteristics o f B ig Data? 4 . W h at p ro ce s s in g te c h n iq u e is ap p lie d to p ro c e s s B i ta?

C hapter 1 * An Overview o f B u sin ess Intelligence, Analytics, and D ecisio n Support 59

Application Case 1.7 G ilt Groupe's Flash Sales Streamlined by Big Data Analytics Gilt G ro u p e is a n o n lin e d estination offering flash sjdes for m ajor brand s b y selling their cloth in g and accessories. It o ffers its m em b ers ex clu siv e discounts on high-end clo th in g an d o th er apparel. After regis- lering w ith Gilt, cu sto m ers are s e n t e-m ails containing i variety o f offers. Custom ers are given a 3 6 -4 8 hou r w ind ow to m ake p u rch ases using th e se offers. T h e re are a b o u t 30 d ifferent sales e a c h day. W h ile a typical d ep artm en t store turns o v er its inventory tw o o r three rimes a year, Gilt d o e s it eigh t to 10 tim es a year. Thus, th ey hav e to m a n a g e th eir inventory extrem ely w ell : r th ey cou ld in cu r extrem ely h ig h inventory costs. In ord er to d o this, analytics softw are d ev elo p e d at Gilt k e e p s track o f every cu sto m e r click — ranging from w h at brand s th e cu stom ers click on , w h at colors ih e y c h o o s e , w h a t styles th e y p ick, an d w h at they en d up buying. T h e n Gilt tries to p red ict w h at th ese cu stom ers are m o re likely to b u y and stocks inven- :o iy accord in g to th e se pred iction s. Custom ers are <ent cu stom ized alerts to sa le offers d ep en d in g o n th e rjg g e s tio n s b y th e analytics softw are.

T h at, h o w e v e r, is n o t th e w h o le p ro ce ss. T h e softw are also m o n ito rs w h at offers th e cu stom ers c h o o s e fro m th e re co m m e n d e d offers to m ak e m o re accu rate p re d ictio n s an d to in cre a se th e e ffectiv en ess o f its p e rso n a liz e d reco m m en d atio n s. S o m e cu sto m ­ ers d o n o t c h e c k e-m ail that often . G ilt’s analytics

softw are k e e p s track o f re s p o n s e s to offers an d sen ds th e sa m e o ffe r 3 days la ter to th o se cu stom ers w h o h a v en ’t resp on d ed . G ilt a lso k e e p s track o f w h at cu sto m ers are saying in g e n eral a b o u t Gilt’s p ro d ­ u cts b y analyzing T w itte r fe e d s to an alyze sen tim ent. Gilt’s re co m m en d atio n so ftw are is b a se d o n T erad ata Aster’s te ch n o lo g y so lu tio n that in clu d es B ig D ata analytics te ch n o lo g ie s.

Q u e s t i o n s f o r D i s c u s s i o n

1. W h a t m a k e s this c a s e study a n e x a m p le o f B ig D ata analytics?

2. W h a t ty p e s o f d e c is io n s d o e s G ilt G ro u p e h av e to m ake?

W h a t W e C a n L e a m F r o m t h i s A p p lic a tio n C a se

T h e re is co n tin u ou s g ro w th in th e am o u n t o f struc­ tu red an d unstructured data, and m any organiza­ tio n s are n o w tapping th e s e data to m ak e action ab le d ecisio n s. B ig D ata analytics is n o w e n a b led b y the ad van cem en ts in te ch n o lo g ie s that aid in storag e and p ro cessin g o f vast am o u n ts o f rapidly grow ing data.

Source-. Asterdata.com, “Gilt Groupe Speaks on Digital Marketing Optimization," asterdata.com/gilt_groupe_video.php (accessed February 2013).

1.10 PLAN OF THE BOOK T h e previou s s e c tio n s h a v e g iv e n y o u a n u n d erstan d in g o f th e n e e d fo r u sin g in form a­ n t te c h n o lo g y in d e c is io n m ak in g ; a n IT -o rie n te d v ie w o f vario u s ty p e s o f d e c is io n s ; i o d th e e v o lu tio n o f d e c is io n su p p o rt sy stem s in to b u s in e s s in te llig e n c e , a n d n o w in to m a ly tics. In th e la st tw o s e c tio n s w e h a v e s e e n a n o v erv iew o f v ario u s ty p e s o f an aly t­ i c an d th e ir a p p lica tio n s. N ow w e a re re ad y fo r a m o re d eta iled m a n a g e ria l e x c u rs io n e e o th e se to p ic s , a lo n g w ith s o m e p o ten tia lly d e e p h a n d s -o n e x p e r ie n c e in s o m e o f the Technical to p ics. T h e 1 4 c h a p te rs o f this b o o k are o rg an ize d in to fiv e parts, as s h o w n in

fig u r e 1.6.

=art l: Business A nalytics: An O verview fc. Chapter 1, w e p ro v id e d a n in trod u ctio n , d efin ition s, a n d a n o v erv iew o f d e c is io n sup- r c t t sy stem s, b u s in e s s in te llig e n ce , an d analytics, in clu d in g B ig D ata analytics. C h ap te r 2

rvers th e b a sic p h a s e s o f th e d ecisio n -m a k in g p ro c e s s a n d in trod u ces d e c is io n su p p ort

553cems in m o re d etail.

6 0 Part I • D ec isio n Making and Analytics: An Overview

P a r t I D e cisio n M a k in g a n d A n a ly tic s : A n O verview

C h a p t e r 1 An Overview of Business

Intelligence, Analytics, and Decision Support

C h a p t e r 2 Foundations and Technologies for

Decision Making

1

P a r t III P a r t IV P re d ictiv e A n a ly t ic s P r e s c r ip t iv e A n a ly t ic s

P a r t II D e s c r ip t iv e A nalytic

C h a p t e r 3 Data Warehousing

C h a p t e r 4 Business Reporting. Visual

Analytics, and Business Performance Management

C h a p t e r 5 Data Mining

C h a p t e r 6 Techniques for Predictive

Modeling

C h a p t e r 7 Text Analytics, Text Mining, and

Sentiment Analysis

C h a p t e r S W eb Analytics, W eb Mining, and

Social Analytics

C h a p t e r 9 Model-Based Decision Making: Optimization and Multi-Criteria

Systems

C h a p t e r 1 0 Modeling and Analysis:

Heuristic Search Methods and Simulation

C h a p t e r 1 1 Automated Decision Systems and

Expert Systems

C h a p t e r 1 2 Knowledge Management and

Collaborative System s

P a r t V B ig D a ta a n d F u t u r e D ire c tio n s

f o r B u s in e s s A n a ly tic s

C h a p t e r 1 3 Big Data and Analytics

C h a p t e r 1 4 Business Analytics: Emerging Trends and Future Impacts

P a r t V I Online S u p p le m e n ts

FIGURE 1.6 Plan of the Book.

Part II: Descriptive Analytics P art II b e g in s w ith a n in trod u ctio n to d ata w a re h o u sin g issu es, ap p licatio n s, and te c h n o lo ­ g ie s in C h ap ter 3. D ata re p re se n t th e fu nd am ental b a c k b o n e o f an y d e c is io n su p p o rt and analytics ap p licatio n . C h ap ter 4 d e s crib e s b u s in e s s rep ortin g, v isu alization te ch n o lo g ie s, a n d ap p licatio n s. It a lso in clu d es a b rie f o v erv iew o f b u sin e ss p e rfo rm a n ce m an ag em en t te ch n iq u e s a n d ap p licatio n s, a to p ic th a t h a s b e e n a k e y part o f traditional B I.

Part III: Predictive A nalytics P art III co m p rise s a large part o f th e b o o k . It b e g in s w ith a n in trod u ctio n to p red ictive analytics a p p lica tio n s in C h ap ter 5- It in clu d es m a n y o f th e co m m o n a p p lica tio n te ch ­ n iq u es: cla ssificatio n , clu sterin g, a s s o cia tio n m ining , a n d so fo rth. C h ap ter 6 in clu d es a te ch n ica l d escrip tio n o f s e le c te d d ata m in in g te ch n iq u e s , e sp e cia lly n eu ral n e tw o rk m o d ­ e ls. C h ap ter 7 fo c u s e s o n te x t m in in g a p p lica tio n s. Sim ilarly, C h ap ter 8 fo c u s e s o n W e b analytics, in clu d in g so cia l m ed ia analytics, s en tim en t analysis, a n d o th e r re la ted to p ics.

Chapter 1 * An Overview o f B u sin ess Intelligence, Analytics, and D ec isio n Support 61

Part IV: Prescrip tive Analytics Part IV in tro d u ces d e c is io n an alytic te ch n iq u e s, w h ic h a re a lso calle d p rescrip tiv e analyt­ ics. S p ecifically , C h ap te r 9 co v ers s e le c te d m o d els that m ay b e im p lem e n te d in sp read ­ sh e et e n v iro n m en ts. It a lso c o v e rs a p o p u la r m u lti-o b jectiv e d e cisio n te c h n iq u e analytic

hierarch y p ro ce s s e s. C h ap ter 10 th e n in tro d u ces o th er m o d e l-b a s e d d ecisio n -m a k in g te c h n iq u e s , e sp e ­

cially heu ristic m o d e ls an d sim ulation. C h ap ter 11 in tro d u ces au to m ate d d e c is io n system s inclu ding e x p e rt system s. T h is p a n c o n c lu d e s w ith a b r ie f d iscu ssion o f k n o w le d g e m a n a g e m e n t a n d g ro u p su p p o rt system s in C h ap ter 12.

Part V: Big Data and Future Directions fo r Business A nalytics Part V b e g in s w ith a m o re d eta iled co v e ra g e o f B ig D ata and analytics in C h ap te r 13-

C h a p te r 1 4 a tte m p ts to in te g ra te all th e m ate rial c o v e r e d in th is b o o k and c o n c lu d e s w ith a d is c u s s io n o f e m e rg in g tre n d s, s u c h a s h o w th e u b iq u ity o f w ir e ­ le s s an d G P S d e v ic e s an d o th e r s e n s o r s is re s u ltin g in th e c r e a tio n o f m a ssiv e n e w d a ta b a s e s a n d u n iq u e a p p lic a tio n s . A n e w b r e e d o f d ata m in in g an d B I c o m p a n ie s is e m e rg in g to a n a ly z e th e s e n e w d a ta b a s e s a n d c r e a te a m u c h b e tte r a n d d e e p e r u n d e r­ stan d in g o f c u s to m e r s ’ b e h a v io rs a n d m o v e m e n ts . T h e c h a p te r a ls o c o v e r s c lo u d -b a s e d an a ly tics, r e c o m m e n d a tio n sy stem s, a n d a b r ie f d is c u ss io n o f secu rity / p riv acy d im e n ­ sio n s o f a n a ly tic s . It c o n c lu d e s th e b o o k b y a ls o p re s e n tin g a d is c u s s io n o f th e a n a ly tics e c o s y s te m . A n u n d e rs ta n d in g o f th e e c o s y s te m a n d th e v a rio u s p la y e rs in th e a n a ly tics in d u stry h ig h lig h ts th e v a rio u s c a r e e r o p p o rtu n itie s fo r stu d e n ts an d p ra c titio n e rs o f

an aly tics.

1.11 RESO U RCES, LIN K S, A N D THE T ER A D A T A U N IVERSITY NETW ORK CONNECTION

T h e u s e o f this c h a p te r an d m o st o th er ch ap ters in this b o o k c a n b e e n h a n c e d b y th e to o ls d e s crib e d in th e fo llo w in g sectio n s.

Resources and Links W e re co m m e n d th e fo llo w in g m a jo r re so u rce s an d links:

• T h e D ata W a re h o u sin g Institute (tdw i.org) • In fo rm a tio n M an ag em en t (rnform ation-m anagem ent.com ) • D SS R e so u rc e s (d ssresou rces.co m ) • M icroso ft E n terp rise C o n sortiu m ( e n t e r p r i s e . w a l t o n c o l l e g e . u a r k . e d u / m e c . a s p )

Vendors, Products, and Demos M ost v e n d o rs p ro v id e so ftw are d em o s o f th e ir p ro d u cts an d ap p licatio n s. In fo rm ation a b o u t p ro d u cts, arch ite ctu re, a n d so ftw are is a v a ilab le at d ssresou rces.co m .

Periodicals W e re co m m en d th e fo llo w in g p eriod icals:

• D ecision S u pport System s • CIO In sight (cioinsight.com ) • T echn ology E v alu ation (technologyevaluation.com ) • B a selin e M ag azin e (baselinem ag.com )

6 2 Part I • D ecisio n M aking and Analytics: An Overview

The Teradata U niversity N etw ork Connection T h is b o o k is tightly c o n n e c te d w ith th e fr e e r e s o u r c e s p ro v id e d b y T e ra d a ta U n iv ersity N e tw o rk (T U N ; s e e te ra d a ta u n iv e rsity n e tw o rk .co m ). T h e TU N p o rta l is d ivid ed in to tw o m a jo r p a rts: o n e fo r stu d e n ts a n d o n e fo r fa cu lty . T h is b o o k is c o n n e c te d to th e TU N p o rta l v ia a s p e c ia l s e c tio n a t th e e n d o f e a c h c h a p te r. T h a t s e c tio n in clu d e s a p p r o p ria te lin k s fo r th e s p e c ific c h a p te r, p o in tin g to re le v a n t re s o u r c e s . In ad d ition, w e p ro v id e h a n d s -o n e x e r c is e s , u sin g s o ftw a re a n d o th e r m a te ria l (e .g ., c a s e s ) avail­ a b le a t TUN.

The Book's W eb Site T h is b o o k ’s W e b site, pearson glob aled ition s.com /tu rb an , co n ta in s su p p le m e n tal te x ­ tual m aterial o rg an ize d as W e b ch ap ters th at co r re s p o n d to th e p rin ted b o o k ’s chapters. T h e to p ics o f th e s e ch ap ters are listed in th e o n lin e ch a p te r ta b le o f co n ten ts. O th e r c o n ­ te n t is a lso a v ailab le o n a n in d e p e n d e n t W e b site (d ssbibook.com ).2

Chapter Highlights

• T h e b u s in e s s e n v iro n m en t is b e c o m in g c o m p le x an d is rap id ly ch a n g in g , m ak in g d e c is io n m aking m o re difficult.

• B u s in e s s e s m u st re sp o n d an d ad ap t to th e ch a n g ­ ing e n v iro n m en t rap id ly b y m ak in g faste r an d b e tte r d ecisio n s.

• T h e tim e fram e fo r m ak in g d ecisio n s is shrinkin g, w h e re a s th e g lo b a l n atu re o f d e cisio n m ak in g is e x p a n d in g , n e cessitatin g th e d ev e lo p m e n t an d u s e o f co m p u te rize d DSS.

• C o m p u te rize d su p p o rt fo r m an ag ers is o ften e sse n tia l fo r th e survival o f a n organization.

• An e a rly d e cisio n su p p ort fram ew ork divides d e c is io n situations into n in e categ o ries, d ep en d in g o n th e d e g re e o f stru ctured ness an d m anagerial activities. E a ch c a te g o iy is su p p orted differently.

• Structured repetitive d ecisio n s are su p p orted b y standard quantitative analysis m ethod s, su ch a s MS, MIS, an d ru le-b ased autom ated d ecisio n support.

• D SS u s e data, m o d els, and s o m e tim e s k n o w le d g e m an a g e m e n t to find so lu tio n s fo r sem istm ctu re d a n d s o m e u n stru ctu red p ro b lem s.

• B I m e th o d s utilize a cen tral rep ository calle d a d ata w a re h o u s e th at e n a b le s e fficie n t data m ining, OLAP, BPM , and d ata visualization.

• B I a rch ite ctu re in clu d e s a d ata w a reh o u se , b u si­ n e ss an aly tics to o ls u s e d b y e n d u sers, a n d a user in te rface (s u c h as a d ash b o ard ).

• M any o rg an izatio n s e m p lo y d escrip tiv e analytics to re p la c e th e ir traditional flat rep ortin g w ith inter­ activ e re p o rtin g that p ro v id es insights, trend s, and p attern s in th e tran saction al data.

• P red ictiv e an aly tics e n a b le o rg an ization s to estab ­ lish p re d ictiv e ru les th at drive th e b u sin ess ou t­ c o m e s th ro u g h h istorical data analysis o f the existin g b e h a v io r o f th e cu stom ers.

• P rescrip tive an alytics h e lp in b u ild in g m o d e ls that involve fo re ca s tin g a n d o p tim ization te ch n iq u es b a s e d o n th e p rin cip le s o f o p era tio n s research an d m an a g e m e n t s c ie n c e to h e lp o rg an ization s to m a k e b e tte r d ecisio n s.

• B ig D ata an aly tics fo c u s e s o n un structured , large d ata sets th a t m ay a lso in clu d e vastly different typ es o f d ata fo r analysis.

• A nalytics as a field is a lso k n o w n b y industry- s p e cific a p p lica tio n n a m e s s u ch as sp orts analytics. It is a lso k n o w n b y o th e r related n a m e s s u ch as d ata s c ie n c e o r n e tw o rk s cie n ce .

1 As this book went to press, we verified that all the cited Web sites were active and valid. However, URLs are dynamic. Web sites to which we refer in the text sometimes change or are discontinued because companies change names, are bought or sold, merge, or fail. Sometimes Web sites are down for maintenance, repair, or redesign. Many organizations have dropped the initial “www” designation for their sites, but some still use it. If you have a problem connecting to a Web site that we mention, please be patient and simply run a Web search to try to identify the possible new site. Most times, you can quickly find the new site through one of the popular search engines. We apologize in advance for this inconvenience.

Chapter 1 • An O verview o f B u sin ess Intelligen ce, Analytics, and D ecisio n Support 6 3

Key Terms b u sin ess in te llig e n ce

( B l ) d ash bo ard d ata m in in g

d e c is io n (o r n orm ativ e) analytics

d e c is io n su p p o rt system (D S S )

d escrip tiv e ( o r rep ortin g) analytics

p red ictiv e an alytics prescrip tiv e an alytics

sem istru ctu red p ro b le m

stru ctu red p ro b le m unstructured p ro b lem

Questions for Discussion 1 . D istinguish betw een strategic and tactical planning? 2 . W hat is data mining and why is it classified under predictive

analytics? Search the W eb for an exam ple o f data mining in an organization o f you r ch o ice and illustrate th e w ay it is currently in use.

3 . Prescriptive analytics is considered to b e a step further ahead o f predictive analysis and substantially different

from it. Provide an exam ple o f ea ch and outline their differences.

4 . Provide a definition o f B l. 5 . D efine m anagerial decision making. Discuss this con cep t in

the context o f the four-step approach to d ecision making.

Exercises T e ra d a ta U n iv e rs ity N e tw o r k (T U N ) a n d O th e r H an d s-O n E x e r c i s e s 1. G o to te ra d a ta u n iv e r s ity n e tw o r k .c o m . Using the reg­

istration you r instructor provides, log o n and learn the con tent o f th e site. Y o u will receiv e assignm ents related to this site. Prepare a list o f 2 0 item s in th e site that you think cou ld b e b en eficial to you.

2 . Enter th e TUN site and sele c t “cases, projects and assign­ m en ts.” T h e n select the c a se study: “Harrah’s High Payoff from C ustom er Inform ation.” Answ er th e follow ing ques­ tions ab o u t this case: a . W hat inform ation d oes th e data m ining generate? b . H ow is this inform ation helpful to m anagem ent in

d ecision making? (B e specific.) c . List the types o f data that are mined. d . Is this a DSS o r B l application? Why?

? . G o to te ra d a ta u n iv e rsity n e tw o rk .c o m and find the paper titled “Data Warehousing Supports Corporate Strategy at First American Corporation” (by Watson, W ixom, and Goodhue). Read th e paper and answer the following questions: a . W hat w ere the drivers for the DW/BI p roject in the

com pany? b . W hat strategic advantages w ere realized? c . W hat operational and tactical advantages w ere achieved? d . W hat w ere the critical su ccess factors (CSF) for the

im plementation? -i, G o to a n a ly tic s-m a g a z in e .o rg / is su e s/ d ig ita l-e d itio n s

and find the January/February 2012 edition titled “Special Issue: T h e Future o f Healthcare.” Read the article “Predictive

Analytics— Saving Lives and Lowering Medical Bills.” Answer th e following questions: a . W hat is th e problem that is b ein g addressed b y apply­

ing predictive analytics? b . W hat is the FICO M edication A dh erence Score? c . H ow is a prediction m od el trained to predict the FICO

M edication A dherence Score? D id the prediction m od el classify FICO M edication A dh erence Score?

d. Z oom in o n Figure 4 and exp lain w hat kind o f tech­ niqu e is applied o n th e generated results.

e . List so m e o f th e action able d ecisions that w ere based o n th e results o f th e predictions.

5 . Visit h ttp ://w w w .ib m .c o m /a n a ly tic s /u s /e n /w h a t-is - s m a rte r -a n a ly tics /b ig -d a ta -a n a ly sis .h tm l. Read the s e c ­ tion “Gain actionable insights from big data analysis,” and w atch th e video “See how analytics can help organizations in all industries use big data to achieve significant outcom es” under Case Studies to answer the following questions: a . E xplain b ig d a t a and its im portance in d ecision m ak­

ing w ith exam ples. b . A ppraise the m axim “w ithout analytics, b ig data is just

n o ise .” c . In w h ic h sectors and areas might big data analytics b e

m ost useful? G ive exam ples. d . W hat are the suggested solutions to m anaging risks? e . Review and analyze a c a se study from IBM ’s W eb site

and exp lain h ow b ig data analytics provided solutions. 6 . B usiness analytics and com puterized data processing

support m anagers and d ecision m aking. K eep in g current

6 4 Part I • D ecisio n Making and Analytics: An O verview

bu sin ess environm en t challeng es in mind, along with M intzberg’s 10 m anagerial roles ( s e e T a b l e 1 .2 ), identify five su ch roles in com panies/governm ent’s press release and com m unications.

7 - G o to oracle.co m , a leading com p an y in B I. M ake a map o f their W eb site illustrating their products and com m uni­ cation styles w ith available resou rces fo r business.

8 . Search th e W eb fo r a com pany that uses the four m ajor com p on en ts o f a standard B I system.

9 . In the com p an y identified in th e previous question, illus­ trate their m ain prod ucts and style o f B I and discuss the m ain tools used. R efer to th e tools m en tioned in this chapter.

End-of-Chapter Application Case

Nationwide Insurance Used BI to Enhance Customer Service Nationwide Mutual Insurance Company, headquartered in Columbus, O hio, is o n e o f the largest insurance and financial services com panies, with $23 billion in revenues and m ore than $160 billion in statutory assets. It offers a com prehensive range o f products through its family o f 100-plus com panies with insurance products fo r auto, m otorcycle, boat, life, hom eow n­ ers, and farms. It also offers financial products and services including annuities, mortgages, mutual funds, pensions, and investment m anagement.

Nationwide strives to achieve greater efficiency in all operations by m anaging its ex p en ses along with its ability to grow its revenue. It recognizes the u se o f its strategic asset o f information com bin ed with analytics to ou tp ace com petitors in strategic and operational decision making even in com plex and unpredictable environments.

Historically, Nationwide’s business units w orked inde­ pendently and with a lot o f autonom y. This led to duplication o f efforts, widely dissimilar data processing environments, and extrem e data redundancy, resulting in higher expen ses. The situation got com plicated w hen Nationwide pursued any merg­ ers o r acquisitions.

Nationwide, using enterprise data w arehouse technology from Teradata, set ou t to create, from the ground up, a single, authoritative environment for clean, consistent, and com plete data that can b e effectively used for best-practice analytics to m ake strategic and tactical business decisions in the areas o f custom er growth, retention, product profitability, cost contain­ m ent, and productivity improvements. Nationwide transformed its siloed business units, which w ere supported by stove-piped data environments, into integrated units by using cutting-edge analytics that w ork with clear, consolidated data from all o f its business units. T h e Teradata data w arehouse at Nationwide has grown from 400 gigabytes to more than 100 terabytes and supports 85 percent o f Nationwide’s business with m ore than 2,500 users.

I n t e g r a t e d C u s to m e r K n o w le d g e N ationw ide’s C u stom er K n ow led g e Store (C K S) initiative d ev elo p ed a cu stom er-cen tric d atabase that integrated cu s­ tom er, prod uct, a n d extern ally acq u ired data from m ore

th an 4 8 sou rces into a sin g le cu stom er data m art to deliver a holistic v iew o f cu stom ers. This data m art w as c o u p led with T erad ata’s cu stom er relatio n sh ip m an agem en t ap p lication to c reate and m an age effe ctiv e cu stom er m arketing cam paigns that u s e behavioral analysis o f cu stom er interactions to drive cu stom er m an agem en t action s (CMAs) for target segm ents. N ationw ide ad ded m o re sop h isticated cu stom er analytics that lo o k e d at cu stom er p ortfolios and th e effectiv en ess o f various m arketing cam p aign s. T h is data analysis help ed N ationw ide to initiate p roactive cu stom er com m u nications around cu stom er lifetim e ev en ts like m arriage, birth o f child, or h o m e p u rch ase a n d had significant im p act o n im prov­ ing cu stom er satisfaction. Also, b y integrating cu stom er co n ta ct history, p rod u ct o w nership , and p ay m ent inform a­ tion , N ationw ide’s beh av ioral analytics team s further created prioritized m odels th a t cou ld identify w h ic h sp ecific cu s­ to m er in teraction w as im p ortan t for a custom er at an y given time. This resulted in o n e p ercen ta g e p oint im provem en t in cu stom er reten tio n rates and sig nificant im provem ent in cu stom er enth u siasm sco res. N ationw ide also ach ieved 3 p ercen t ann ual grow th in increm ental s a le s by using CKS. T h e re are o th er uses o f the cu stom er d atabase. In o n e o f th e initiatives, by integrating cu stom er te le p h o n e data from m ultiple system s into CKS, th e relation ship m an agers at N ationw ide tiy to b e proactives in con tactin g custom ers in ad van ce o f a p ossib le w ea th er catastrop h e, su ch as a hur­ rican e o r flood , to p rov id e th e prim aiy p o licy h o ld er infor­ m ation and exp lain th e claim s p ro c esses. T h e s e and o th er analytic insights n o w drive N ationw ide to provide extrem ely person al cu stom er service.

F in a n c ia l O p e r a tio n s A similar perform ance p ay off from integrated information was also noted in financial operations. Nationwide’s decentralized m anagem ent style resulted in a fragmented financial report­ ing environm ent that included m ore than 14 general ledgers, 20 charts o f accounts, 17 separate data repositories, 12 different reporting tools, and hundreds o f thousands o f spreadsheets. Th ere w as n o com m o n central view o f th e business, w hich resulted in labor-intensive slow and inaccurate reporting.

Chapter 1 * An O verview o f B u sin ess In tellig en ce, Analytics, and D ecision Support 6 5

About 75 p ercen t o f th e effort was spent o n acquiring, clean­ ing, and consolidating and validating the data, and very7 little time was spent o n meaningful analysis o f the data.

T h e Financial P erform an ce M anagem ent initiative im plem ented a n e w operating ap p roach that w o rk ed o n a single data and tech n o log y architecture with a com m on set o f system s standardizing th e p rocess o f reporting. It enabled Nationwide to o p erate analytical cen ters o f e x c e lle n c e with w orld-class planning, capital m anagem ent, risk assessm ent, ttA other d ecisio n support capabilities that delivered timely, accurate, and efficien t accounting, reporting, and analytical services.

T h e data fro m m ore than 2 0 0 o p eratio n al system s was sent to the en terp rise-w id e data w areh ou se and th e n distrib­ uted to various ap p lications and analytics. This resulted in a 5 0 p e rc e n t im p rovem en t in the m onthly closin g p ro cess w ith clo sin g intervals red u ced from 14 days to 7 days.

P o s t m e r g e r D a t a I n t e g r a t i o n N ationw id e’s G o al State R ate M anag em ent initiative e n a ­ bled th e c o m p a n y to m erge A llied In su ra n c e ’s au to m o b ile c o iic v system in to its existin g system . B o th N ationw id e and Allied s o u rc e system s w e re cu stom -bu ilt ap p lication s that ■did not sh a re a n y c o m m o n v alu es o r p ro c e ss data in th e v.rnp m an ner. N ationw id e’s IT d ep artm en t d ecid ed to bring i l l th e data fro m s o u rc e system s in to a cen tralized data « ^ re h o u se , o rg an ized in a n in tegrated fash io n that resulted i a standard d im en sio n a l rep orting and h e lp e d N ationw ide

p erform in g w h a t-if an aly ses. T h e data analysis team could identify p rev iou sly u n k n o w n p o ten tia l d ifferen ces m th e data en v iro n m en t w h e r e prem ium s rates w e re c a l­ culated d ifferently b e tw e e n N ationw id e and A llied sides. Correcting all o f th e s e b e n e fite d N ationw id e’s p o licy h o ld ­ ers b e c a u se th e y w e re safeg u ard ed from e x p e rie n c in g w ide prem ium rate sw in gs.

E n h a n c e d R e p o r t i n g ,*2C«3nwide’s legacy reporting system , w hich catered to the c c e c s o f property and casualty business units, to o k w eek s D com pile and deliver th e n eed ed reports to th e agents. X ero n w id e determ ined that it n eed ed better access to sales n-k-T policy inform ation to reach its sales targets. It ch o se a

single data w areh ou se ap p roach and, after careful assessm ent o f th e need s o f s a le s m anagem ent and individual agents, selected a bu sin ess intelligence platform that w ould integrate d ynam ic enterp rise dashboards into its reporting systems, m aking it easy fo r th e agents and associates to view policy inform ation at a g la n ce. T h e n e w reporting s y s tem , dubbed R evenue C on n ection , also enabled users to analyze th e infor­ m ation w ith a lo t o f interactive and drill-dow n-to-details cap a­ bilities at various lev els that elim inated the n e e d to gen erate custom ad h o c reports. Revenue C on nection virtually elim i­ nated requ ests for manual policy audits, resulting in huge savings in tim e and m on ey for the business and tech nology team s. T h e reports w ere produced in 4 to 45 secon d s, rather than days o r w eek s, and productivity in som e units improved b y 2 0 to 3 0 percent.

Q u e s t i o n s f o r D i s c u s s i o n

1 . W hy did Nationw ide n eed an enterprise-w ide data w arehouse?

2 . H ow did integrated data drive th e business value? 3 . W h at fo r m s o f a n a ly tics a r e em ployed af Nationwide? 4 . W ith integrated data available in an enterprise data

w areh ou se, w hat other applications cou ld Nationwide potentially develop?

W h a t W e C an L e a r n f r o m T h is A p p lic a tio n C a se T h e p ro p er u se o f integrated inform ation in org aniza­ tion s c a n h e lp a c h iev e b e tte r b u sin ess o u tc o m e s . Many org an ization s n o w rely o n data w areh ou sin g tech n o lo g ies to perform th e o n lin e analytical p ro c e ss e s o n th e data to d erive v a lu a b le insig hts. T h e insights a re used to d evelop pred ictive m o d els that further e n a b le th e grow th o f th e org an ization s by m o re p re c is e ly a ssessin g cu stom er needs. In creasin gly , o rg an izatio n s a re m oving tow ard deriving v alu e from an aly tical ap p licatio n s in real tim e w ith the h e lp o f integrated data from real-tim e data w areh ou sin g te ch n o lo g ie s .

Source: Teradata.com, “Nationwide, Delivering an On Your Side Experience," t e r a d a t a . c o m / W o r k A r e a / l i n k i t . a s p x P L i n k I d e n t i f i e r= id & Ite m ID = l4 7 l4 (accessed February 2013).

References Anehony, R. N. (1965). P la n n in g a n d C o n tro l Systems: A

F r a m e w o r k f o r A n alysis. Cambridge, MA: Harvard University Graduate Sch ool o f Business.

Aaerdata.com . “Gilt G roupe Speaks on Digital Marketing Optimization. ” w w w .a s te r d a ta .c o m /g ilt_ g ro u p e _ v id e o . p h p (accessed February 2013).

Barabasilab.n eu .ed u . “Network S cien ce." b a ra b a s ila b .n e u . e d u /n e tw o r k s c ie n c e b o o k /d o w n lP D F .h tm l (a c cessed February 2013).

B rooks, D . (2 0 0 9 , M ay 18). “In Praise o f D u llness.” N ew Y ork T im es, n y t i m e s .c o t n / 2 0 0 9 / 0 5 / 1 9 / o p i n i o n / 1 9 b r o o k s . h tm l (a c cessed February 2013).

Centers for D isease Control and Prevention, V accines for Children Program , “M odule 6 o f the VFC O perations G uide. ” c d c .g o v / v a c c in e s / p u b s / p in k b o o k / v a c -s to r - a g e .h t m l # s to r a g e (a c cessed Jan u ary 2013).

E ck erson, W. (2 0 0 3 ). S m a r t C o m p a n ie s in t h e 2 1 s t C en tury: T h e S e c re ts o f C r e a t in g S u c c e s s fu l B u s in e s s In te llig en t S o lu tion s. Seattle, WA: T h e D ata W arehousing Institute.

E m c.com . “D ata S cie n c e Revealed: A D ata-Driven G lim pse into th e B u rg eon in g N ew Field .” e m c .c o m / c o lla te ra l/ a b o u t/ n e w s / e m c -d a ta -s c ie n c e -s tu d y -w p .p d f (a c cessed

February 2013). Gorry, G . A., and M. S. Scott-M orton. (1 9 7 1 ). “A Fram ew ork

for M anagem ent Inform ation System s.” S lo a n M a n a g e m e n t R ev iew , Vol. 13, No. 1, pp. 5 5 -7 0 .

INFORMS. “Analytics Section O verview .” in fo r m s .o r g / C o m m u n ity / A n a ly tic s (accessed February 2013).

K een, P. G. W ., and M. S. Scott-M orton. (1 9 7 8 ). D e c is io n S u p p o r t System s: A n O r g a n iz a t io n a l P ers p ectiv e. Reading, MA: Addison-W esley.

Krivda, C. D. (20 0 8 , M arch). “D ialing Up G row th in a Mature Market.” T e r a d a t a M a g a z in e , pp. 1-3-

Magpiesensing.com . “MagpieSensing Cold Chain Analyt­ ics and Monitoring.” m a g p ie se n s in g .c o m / w p -c o n te n t/ uploads/2013/01/ColdChainAnalyticsMagpieSensing- W h i t e p a p e r .p d f (accessed Janu ary 2013).

Mintzberg, H. A. (1 9 8 0 ). T h e N a tu r e o f M a n a g e r ia l W ork. E n glew ood Cliffs, NJ: P rentice Hall.

Mintzberg, H. A. (1 9 9 3 ). T h e R ise a n d F a l l o f S trateg ic P la n n in g . N ew Y o rk : T h e Free Press.

Sim on, H. (1 9 7 7 ). T h e N ew S c ie n c e o f M a n a g e m e n t D ecisio n . E n glew ood Cliffs, NJ: Prentice Hall.

6 6 Part I • D ec isio n M aking and Analytics: An O verview

T ableau softw are.com . “Eliminating W aste at Seattle Chil­ dren’s .” t a b le a u s o ftw a r e .c o m / e lim in a tin g -w a s te -a t- s e a t t le - c h il d r e n s (a c ce s se d February 2013).

Tableausoftw are.com . “K aleida Health Finds Efficiencies, Stays Com petitive.” ta b le a u s o ftw a r e .c o m / le a m / s to r ie s / u s e r - e x p e r ie n c e -s p e e d -th o u g h t-k a le id a -h e a lth (accessed

February 2013). T eradata.com . “Nationw ide, Delivering an O n Y ou r Side

E xp erien ce. te r a d a ta .c o m / c a s e -s tu d ie s / d eliv e r in g -o n - y o u r -s id e -e x p e r ie n c e (a c cessed February 2013).

Teradata.com . “Sabre Airline Solu tions.” tera d a ta .eo m / t/ c a s e - s t u d i e s / S a b r e - A i r l i n e - S o l u t i o n s - E B 6 2 8 1

(a c cessed February 2 013). W ang, X ., et al. (20 1 2 , January/February). “B ranch Reconfigu­

ration P ractice T hrou gh O perations R esearch in Industrial and Com m ercial B a n k o f China.” I n t e r fa c e s , Vol. 42, No. 1,

pp. 3 3 -^ 4 . W atson, H. (2005, W inter). “Sorting Out W hat’s New in Decision

Support.” B u sin ess In te llig e n c e Jo u r n a l. W ikipedia. “O n B a se P ercen tag e.” e n .w ik ip e d ia .o rg / w ik i/

O n _ b a s e _ p e r c e n t a g e (a c ce s se d Janu ary 2013). W ikip ed ia. “S a b e rm e tric s.” e n .w ik ip e d ia .o r g / w ik i/

S a b e r m e t r i c s (a c c e s s e d Ja n u a ry 2 0 1 3 ). Zaleski, A. (2 0 1 2 ). “M agpie Analytics System Tracks Cold-

Chain Products to K eep V accin es, Reagents Fresh.” T e c h n i c a l ly B a lt i m o r e .c o m (a c cessed February 2013).

Zaman, M. (2009, April). “Business Intelligence: Its Ins and Outs.’' te c lin o lo g y e v a lu a tio n .c o m (accessed Febniary 2013).

Ziam a, A., and J . K ash er. (2 0 0 4 ). “D ata M ining Prim er for th e Data W a reh o u sin g P ro fe ss io n a l.” D ay to n , OH: T e r a d a t a .

Foundations and Technolog for Decision Making

LEARNING OBJECTIVES

■ U n d erstan d th e c o n c e p tu a l fo u n d atio n s o f d e c is io n m aking

■ U n d erstand S im o n ’s fo u r p h a s e s o f d e c is io n m akin g: in te llig e n ce , d esign, c h o ic e , a n d im p lem en tatio n

■ U n d erstan d th e esse n tial d efinition o f D SS

■ U nderstand im portant D SS classifications

* L earn h o w D SS su p p o rt fo r d e cisio n m ak in g c a n b e p ro v id e d in p ra ctice

* U n d erstan d D SS c o m p o n e n ts and h o w th ey integrate

Ou r m a jo r fo cu s in this b o o k is th e su p p o rt o f d e c is io n m ak in g throu gh c o m p u te r-b a s e d in form ation sy stem s. T h e p u rp o se o f this c h a p te r is to d e scrib e th e c o n c e p tu a l fo u n d atio n s o f d e cisio n m ak in g an d h o w d e c is io n su p p o rt is p rovided. T h is ch a p te r in clu d e s th e fo llo w in g sectio n s:

2 .1 O p e n in g V ig n e tte : D e c is io n M o d e lin g a t H P U s in g S p r e a d s h e e ts 6 8 2 .2 D e c i s io n M ak in g : In tr o d u c tio n a n d D e fin itio n s 7 0 2 .3 P h a s e s o f th e D e c is io n -M a k in g P r o c e s s 7 2 2 . 4 D e c i s io n M ak in g : T h e I n te llig e n c e P h a s e 7 4 2 .5 D e c i s io n M ak in g : T h e D e s ig n P h a s e 7 7 2 .6 D e c i s io n M ak in g : T h e C h o ic e P h a s e 8 5 2 .7 D e c i s io n M ak in g : T h e I m p le m e n ta tio n P h a s e 8 5 2 . 8 H o w D e c is io n s A re S u p p o rte d 8 6 2 . 9 D e c i s io n S u p p o rt S y s te m s : C a p a b ilitie s 8 9

2 .1 0 D S S C la s s ific a tio n s 91 2 .1 1 C o m p o n e n ts o f D e c is io n S u p p o r t S y s te m s 9 4

6 8 Part I • D ec isio n M aking and Analytics: An O verview

2.1 OPENING VIGNETTE: Decision Modeling at HP Using Spreadsheets

HP is a m a jo r m an u factu rer o f co m p u ters, p rinters, a n d m an y industrial p ro d u cts. Its vast p ro d u ct lin e lead s to m an y d e c is io n p ro b lem s. O la v s o n an d Fry ( 2 0 0 8 ) h av e w o rk e d o n m an y s p re a d sh ee t m o d e ls fo r assisting d e c is io n m a k e rs at HP a n d h a v e id en tified sev eral le s s o n s fro m b o th th e ir s u c c e s s e s an d th e ir failu res w h e n it co m e s to co n stru ctin g and ap p ly in g sp re a d sh e e t-b a s e d to o ls. T h e y d efin e a to o l as ‘ a re u sab le , analytical so lu tio n d esig n e d to b e h an d e d o f f to n o n te ch n ica l e n d u s e rs to assist th e m in solv in g a re p e a ted

b u sin e ss p ro b le m .” W h e n trying to s o lv e a p ro b lem , H P d e v e lo p e rs co n s id e r th e th ree p h a se s in d ev el­

o p in g a m o d e l. T h e first p h a se is p ro b le m fram ing, w h e re th e y c o n s id e r th e fo llo w in g q u e stio n s in o rd e r to d e v e lo p th e b e s t so lu tio n fo r th e p ro blem :

• W ill an alytics so lv e th e problem ? • C an a n e x istin g so lu tio n b e leveraged ? • Is a to o l need ed ?

T h e first q u e s tio n is im portant b e c a u s e th e p ro b le m m ay n o t b e o f a n an alytic natu re, a n d th e re fo re , a s p re a d sh e e t to o l m ay n o t b e o f m u c h h e lp in th e lo n g ru n w ith o u t fixing th e n o n an aly tical p art o f th e p ro b le m first. F o r e x a m p le , m an y in v en tory -related issu es a rise b e c a u s e o f th e in h e re n t d iffe ren ce s b e tw e e n th e g o a ls o f m arketin g a n d su p p ly c h a in g ro u p s. M arketing lik e s to h av e th e m ax im u m variety in th e p ro d u ct lin e , w h ere a s su p p ly ch a in m a n a g e m e n t fo c u s e s o n red u cin g th e in v en tory co sts. T h is d iffe re n ce is par­ tially ou tsid e th e s c o p e o f an y m o d e l. C o m in g u p w ith n o n m o d e lin g so lu tio n s is im p or­ tan t as w ell. I f th e p ro b le m arises d u e to “m isalig n m en t” o f in ce n tiv e s o r u n c le a r lin es o f authority o r p lan s, n o m o d e l c a n h e lp . T h u s, it is im p ortan t to id entify th e ro o t issue.

T h e s e c o n d q u e s tio n is im portant b e c a u s e so m e tim e s a n existin g to o l m ay so lv e a p ro b le m th a t th e n sav e s tim e a n d m o n ey . S o m etim e s m odifying a n e x istin g to o l m ay solve th e p ro b lem , ag ain saving s o m e tim e an d m o n e y , b u t s o m e tim e s a cu sto m to o l is n e c e s ­ sary to so lv e th e p ro b lem . T h is is clearly w o rth w h ile to e x p lo re .

• T h e third q u e s tio n is im p o rtan t b e c a u s e s o m e tim e s a n e w c o m p u te r-b a s e d sy stem is n o t re q u ire d to s o lv e th e p ro b le m . T h e d e v e lo p e rs h a v e fo u n d th a t th e y o fte n u se an a ly tica lly d eriv ed d e c is io n g u id e lin e s in s te a d o f a to o l. T h is s o lu tio n re q u ire s le s s tim e fo r d e v e lo p m e n t a n d training, h a s lo w e r m a in te n a n c e re q u ire m e n ts, a n d a ls o p ro v id e s s im p le r a n d m o re intuitive resu lts. T h a t is, a fte r th e y h a v e e x p lo r e d th e p ro b le m d e e p e r, th e d e v e lo p e rs m ay d ete rm in e th a t it is b e tte r to p re s e n t d e c is io n ru le s th a t c a n b e e a s ­ ily im p le m e n te d a s g u id e lin e s fo r d e c is io n m a k in g ra th e r th a n a s k in g th e m a n a g e rs to ru n s o m e ty p e o f a c o m p u te r m o d e l. T h is resu lts in e a s ie r train in g , b e tte r u n d e rstan d in g o f th e ru le s b e in g p ro p o s e d , an d in c re a s e d a c c e p ta n c e . It a lso ty p ically le a d s t o lo w e r d e v e lo p m e n t c o s ts a n d re d u c e d tim e fo r d e p lo y m e n t.

I f a m o d e l h a s to b e built, the d e v e lo p e rs m o v e o n to th e s e c o n d p h a s e th e actu al d esig n an d d ev e lo p m e n t o f th e to o ls. A dhering to five g u id elin es te n d s to in c re a se th e p ro b a b ility that th e n e w to o l w ill b e s u cce ssfu l. T h e first g u id elin e is to d e v e lo p a p ro to ­ typ e as q u ick ly as p o ssib le . T h is allow s th e d e v e lo p e rs to te st th e d esign s, d em on strate vario u s fe a tu res a n d id eas fo r th e n e w to o ls, g e t e a rly fe e d b a c k fro m th e e n d u sers to s e e w h a t w o rk s fo r th e m a n d w h a t n e e d s to b e ch a n g e d , a n d te st ad o p tio n . D ev e lo p in g a p ro to ty p e a lso p rev en ts th e d e v e lo p e rs fro m ov erb u ild in g th e to o l an d y e t a llo w s th em to co n stru ct m o re s c a la b le an d stan d ard ized so ftw a re a p p lica tio n s later. A dditionally, b y d ev elo p in g a p ro toty p e, d e v e lo p e rs c a n sto p th e p r o c e s s o n c e th e to o l is “g o o d e n o u g h ,” rath er th a n b u ild in g a stan d ard ized s o lu tio n th a t w o u ld ta k e lo n g e r to b u ild a n d b e m o re

e x p e n s iv e .

C hapter 2 • Foundations and T ech n o lo g ies for D ecision Making

T h e s e c o n d g u id e lin e is t o “b u ild in sig h t, n o t b la c k b o x e s .” T h e H P s p re a d s h e e t a o d e l d e v e lo p e r s b e lie v e th a t th is is im p o rta n t, b e c a u s e o fte n ju s t e n te r in g s o m e d ata

2n d r e c e iv in g a c a lc u la te d o u tp u t is n o t e n o u g h . T h e u se rs n e e d to b e a b le to th in k o f a ltern a tiv e s c e n a r io s , a n d th e to o l d o e s n o t s u p p o rt this i f it is a '‘b la c k b o x ” that p ro v id e s o n ly o n e re c o m m e n d a tio n . T h e y a rg u e th a t a to o l is b e s t o n ly i f it p ro v id e s in fo rm a tio n to h e lp m a k e a n d s u p p o rt d e c is io n s ra th e r th a n ju st g iv e th e a n s w e rs . T h e y lis o b e lie v e th a t a n in te ra c tiv e to o l h e lp s th e u sers to u n d e rs ta n d th e p r o b le m b e tte r, —e r e fo r e le a d in g to m o re in fo r m e d d e c is io n s .

T h e third g u id elin e is to “re m o v e u n n e e d e d co m p le x ity b e fo re h a n d o ff.” T h is is im portant, b e c a u s e a s a to o l b e c o m e s m o re c o m p le x it re q u ires m o re train in g an d e x p e r- ^se. m o re data, and m o re re calib ratio n s. T h e risk o f b u g s and m isu se a ls o in creases. Som etim es it is b e s t to stud y th e p ro b lem , b e g in m o d e lin g a n d analysis, a n d th e n start shaping th e p ro g ram in to a sim p le -to -u se to o l fo r th e e n d user.

T h e fo u rth g u id elin e is to “p artn er w ith e n d u se rs in d isco v ery and d e s ig n .” B y w o rk ­ ing w ith th e e n d u sers th e d e v e lo p e rs g e t a b e tte r fe e l o f th e p ro b le m and a b etter idea o f w h at th e e n d u se rs w ant. It a lso in cre a se s th e e n d u s e rs ’ ab ility to u se an aly tic to ols. T h e e n d u sers a ls o g ain a b e tte r u n d erstan d in g o f th e p ro b le m an d h o w it is so lv e d u sin g th e n e w to ol. A dditionally, in clu d in g th e en d users in th e d ev e lo p m e n t p r o c e s s e n h a n c e s d i l d e cisio n m a k e rs ’ an aly tical k n o w le d g e and cap ab ilities. B y w o rk in g to g e th e r, th e ir k n o w le d g e an d sk ills c o m p le m e n t e a c h o th er in th e fin a l solu tion.

T h e fifth g u id elin e is to “d e v e lo p a n O p e ra tio n s R e se a rch (O R ) c h a m p io n .” B y involv­ ing e n d u se rs in th e d ev e lo p m e n t p ro c e s s , th e d ev elo p e rs cre a te ch a m p io n s fo r th e n e w to o ls w h o th e n g o b a c k to th e ir d ep artm en ts o r co m p a n ie s an d e n c o u ra g e th e ir c o w o rk ­ ers to a c c e p t an d u s e th em . T h e ch a m p io n s are th e n th e e x p e rts o n th e to o ls in th eir areas and ca n th e n h e lp th o se b e in g in tro d u ced to th e n e w to o ls. H aving ch a m p io n s in cre a se s th e p o ssibility th a t th e to o ls w ill b e a d o p te d in to th e b u s in e s s e s su ccessfu lly .

T h e fin al s ta g e is the h an d o ff, w h e n th e final to o ls that pro v id e c o m p le te so lu tio n s a re given to th e b u sin esses. W h e n p lan n in g th e h an d o ff, it is im p o rtan t to a n s w e r th e fo l­ low ing q u e stio n s:

• W h o w ill u s e th e tool? • W h o o w n s th e d ecisio n s that th e to o l will support? • W h o e ls e m u st b e involved? • W h o is re s p o n s ib le fo r m a in te n a n ce a n d e n h a n c e m e n t o f th e tool? • W h e n will th e to o l b e used? • H o w w ill th e u se o f th e to o l fit in w ith o th e r p ro cesse s? • D o e s it c h a n g e th e p ro cesses? • D o e s it g e n e r a te in p u t in to th o se p ro cesse s? • H o w w ill th e to o l im p a ct b u s in e s s p erfo rm an ce? • A re th e e x istin g m etrics su fficien t to rew ard this a s p e c t o f p erfo rm an ce? • H o w sh o u ld th e m etrics and in ce n tiv e s b e c h a n g e d to m ax im ize im p a ct to th e b u si­

n e s s fro m th e to o l an d process?

B y k e e p in g th e s e le s s o n s in m ind , d e v e lo p e rs a n d p ro p o n e n ts o f c o m p u te riz e d d e c i­ s io n su p p o rt in g e n era l a n d sp re a d sh e e t-b a s e d m o d e ls in p articu lar are lik e ly to e n jo y g re a te r su cce ss.

QUESTIONS F O R TH E OPENING VIGNETTE

1 . W h at are s o m e o f th e k e y q u estio n s to b e ask ed in su p p o itin g d ecisio n m aking throu gh DSS?

2 . W h a t g u id elin es c a n b e le a rn e d fro m this v ig n e tte a b o u t d ev elo p in g DSS?

3. W h at le s s o n s sh o u ld b e k e p t in m in d fo r su cce ssfu l m o d e l im plem entation ?

7 0 Part I • D ecisio n Making and Analytics: An O verview

WHAT W E CAN LEARN FROM THIS VIGNETTE

T h is v ig n e tte re la tes to p ro vid ing d e cisio n su p p o rt in a larg e org anization:

• B e f o r e b u ild in g a m o d e l, d e c is io n m a k e rs sh o u ld d ev elo p a g o o d u n d erstan d in g o f th e p ro b le m that n e e d s to b e ad d ressed .

• A m o d el m ay n o t b e n e ce s s a ry to ad dress th e p ro b lem . • B e fo re developing a n e w tool, d ecision m akers shou ld e xp lo re reu se o f existing tools. • T h e g o al o f m o d e l b u ild in g is to g ain b etter in sig h t in to th e p ro b lem , n o t ju st to

g e n e r a te m o re nu m bers. • Im p le m en tatio n p lan s sh o u ld b e d e v e lo p e d a lo n g w ith th e m odel.

Source: Based on T. Olavson and C. Fry, “Spreadsheet Decision-Support Tools: Lessons Learned at Hewlett- Packard,” Interfaces, Vol. 38, No. 4, July/August 2008, pp. 300-310.

2.2 DECISIO N M AKING: INTRODUCTION A N D DEFINITIONS W e are a b o u t to e x a m in e h o w d e c is io n m ak in g is p ra c tic e d an d s o m e o f th e un d erlying th e o r ie s and m o d e ls o f d e c is io n m ak in g . Y o u will a ls o learn a b o u t th e v ario u s traits o f d e c is io n m ak e rs, in clu d in g w h a t ch a ra cte riz e s a g o o d d e c is io n m ak er. K n o w in g this ca n h e lp y o u to u n d erstan d th e ty p e s o f d e c is io n s u p p o rt to o ls th at m an ag e rs c a n u s e to m a k e m o re e ffe c tiv e d e cisio n s. In th e fo llo w in g s e c tio n s , w e d iscu ss v ario u s a s p e c ts o f d e c is io n m akin g.

Characteristics of Decision M aking In ad dition to th e ch aracteristics p re s e n te d in th e o p e n in g v ignette, decision making m ay involve th e fo llow ing:

• G ro u p th in k (i.e ., gro u p m e m b ers a c c e p t th e so lu tio n w ith o u t th in k in g fo r th em ­ s e lv e s ) c a n lea d to b a d d ecisio n s.

• D e c is io n m ak ers are in te reste d in ev alu atin g w h a t-if scen ario s. • E x p e rim e n tatio n w ith a re al sy stem (e .g ., d e v e lo p a s ch e d u le , try it, an d s e e h ow

w e ll it w o rk s ) m ay result in failure. • E x p e rim e n tatio n w ith a re al sy stem is p o ss ib le o n ly fo r o n e s e t o f co n d itio n s a t a

tim e a n d c a n b e disastrous. • C h an g es in th e d e cisio n -m a k in g en v iro n m en t m ay o c c u r co n tin u o u sly , lead in g to

invalid ating assu m p tio n s a b o u t a situ atio n (e .g ., d eliv eries aro u n d h o lid ay tim es m ay in cre a se , requ irin g a d ifferen t v ie w o f th e p ro b le m ).

• C h an ges in the d ecisio n -m a k in g en v iro n m en t m a y a ffect d e c is io n q u ality b y im pos­ in g tim e p ressu re o n the d e c is io n m aker.

• C o lle ctin g in form ation a n d analy zing a p ro b le m ta k e s tim e an d c a n b e e x p e n s iv e . It is d ifficu lt to d eterm in e w h e n to sto p and m a k e a d ecisio n .

• T h e r e m ay n o t b e s u fficie n t in form ation to m a k e a n in tellig en t d ecisio n . • T o o m u ch in fo rm atio n m ay b e a v ailab le (i.e ., in fo rm atio n o v erlo ad ).

T o d eterm in e h o w real d e cisio n m ak ers m a k e d ecisio n s, w e m ust first u n d erstan d the p ro cess and th e im portant issu es in v o lv ed in d e c is io n m aking. T h e n w e c a n understand ap p rop riate m e th o d o lo g ie s fo r assisting d e cisio n m a k e rs a n d th e co n trib u tio n s inform ation system s c a n m ak e. O nly th e n c a n w e d e v e lo p D SS to h e lp d e c is io n m akers.

T h is c h a p te r is o r g a n iz e d b a s e d o n th e th re e k e y w o rd s th at fo rm th e te rm DSS: d e c i s i o n , s u p p o r t , a n d s y s t e m s . A d e c is io n m a k e r s h o u ld n o t sim p ly ap p ly IT to o ls b lin d ly . R ath er, th e d e c is io n m a k e r g e ts s u p p o rt th ro u g h a ra tio n a l a p p r o a c h that

Chapter 2 • Foundations and T ech n o lo g ies for D ecisio n Making

sim p lifies re a lity a n d p ro v id e s a re la tiv e ly q u ic k a n d in e x p e n s iv e m e a n s o f co n s id e rin g v arious a ltern a tiv e c o u rs e s o f a c tio n to arrive a t th e b e s t ( o r at le a s t a v e ry g o o d ) s o lu ­ tio n t o th e p ro b le m .

A W orking D efinition of Decision M aking D ecision m akin g is a p ro c e s s o f ch o o s in g a m o n g m o o r m o re altern ativ e co u rse s o f actio n fo r th e p u rp o s e o f attaining o n e o r m o re g oals. A cco rd in g to Sim o n ( 1 9 7 7 ) , m an a­ g erial d e c is io n m ak in g is sy n o n y m o u s w ith th e en tire m a n a g e m e n t p ro c e s s . C o n sid er the im p ortan t m an ag erial fu n ctio n o f p lan n in g . P lan n in g in v olv es a serie s o f d ecisio n s: W hat sh o u ld b e d o n e? W hen? W here? W hy? How ? B y w hom ? M an ag ers s e t g o a ls , o r plan; h e n ce , p lan n in g im p lie s d e cisio n m akin g. O th e r m an ag erial fu n ctio n s, s u ch as organizing and co n tro llin g , a ls o involve d e c is io n m aking.

Decision-Making Disciplines D e cis io n m ak in g is d irectly in flu e n ced b y sev eral m a jo r d iscip lin es, s o m e o f w h ic h are b eh av io ral a n d s o m e o f w h ic h are scie n tific in natu re. W e m u st b e aw are o f h o w th eir p h ilo so p h ies c a n a ffe c t o u r ability to m ak e d e cis io n s an d pro v id e su p p ort. B eh av io ral d iscip lin es in c lu d e an th ro p o lo g y , law , p h ilo so p h y , p o litical s c ie n c e , p sy ch o lo g y , so cial p sy ch olog y , a n d so cio lo g y . S cien tific d iscip lin e s in clu d e co m p u te r s c ie n c e , d e cisio n analysis, e c o n o m ic s , e n g in e e rin g , th e h ard s c ie n c e s (e .g ., b io lo g y , ch em istry, p hy sics), m an ag e m e n t s cie n ce / o p e ra tio n s re s e a rch , m ath em atics, and statistics.

An im p ortan t ch a ra cteristic o f m an a g e m e n t su p p o rt system s (M SS) is th e ir e m p h a ­ sis o n th e e ffe c tiv e n e s s, o r “g o o d n e s s ,” o f th e d e c is io n p ro d u ce d rath er th a n o n the com p u tatio n al e ffic ie n c y o f o b ta in in g it; this is u su ally a m a jo r c o n c e r n o f a tra n sa ctio n p ro cessin g sy stem . M ost W e b -b a s e d D SS are fo c u s e d o n im p ro vin g d e c is io n e ffe ctiv e n e ss. E fficie n cy m ay b e a by-p rod u ct.

Decision Style and Decision M akers In th e fo llo w in g s e c tio n s , w e e x a m in e th e n o tio n o f d e c is io n style and s p e c ific a sp e cts a b o u t d e c is io n m akers.

DECISION STYLE D e cisio n sty le is th e m a n n er b y w h ich d e cisio n m ak ers th in k an d react to p ro b lem s. T h is in clu d es th e w ay th e y p e rc e iv e a p ro b lem , th eir co g n itiv e re sp o n se s, in d h o w v a lu e s a n d b e lie fs v a ry fro m individual to individual a n d fro m situ atio n to situation. As a resu lt, p e o p le m a k e d e cis io n s in d ifferen t w ay s. A lthough th e re is a g en eral p ro cess o f d e c is io n m ak in g , it is fa r from linear. P e o p le d o n o t fo llo w th e sam e steps of th e p ro c e s s in th e sa m e s e q u e n c e , n o r d o th ey u se all th e step s. F u rth e rm o re, the em phasis, tim e a llo tm e n t, a n d priorities g iv en to e a c h s te p v ary sig nificantly, n o t only from o n e p e rs o n to an o th er, b u t a lso fro m o n e situ ation to th e n e x t. T h e m a n n e r in w h ich m anagers m a k e d e c is io n s (a n d th e w a y th e y in te ract w ith o th er p e o p le ) d e s c rib e s th eir d ecisio n style. B e c a u s e d e c is io n styles d e p e n d o n the facto rs d e s crib e d e arlier, th e re are m any d e c is io n sty les. P erso n ality te m p e ra m e n t tests are o fte n u se d to d ete rm in e d e cisio n sqdes. B e c a u s e th e re are m an y s u ch tests, it is im p ortan t to try to e q u a te th e m in d eter­ m ining d e c is io n style. H o w ev er, th e v arious tests m e a su re so m e w h a t d iffe ren t a s p e cts o f personality, s o th e y c a n n o t b e e q u ated .

R e se a rch e rs h a v e id en tified a n u m b e r o f d ecisio n -m a k in g styles. T h e s e in clu d e h e u ­ ristic and an aly tic styles. O n e c a n a lso d istinguish b e tw e e n a u to cra tic v e rsu s d em o cra tic styles. A n o th er sty le is co n su ltativ e (w ith individuals o r g ro u p s). O f c o u rs e , th e re are m any co m b in a tio n s an d v ariation s o f styles. F o r e x a m p le , a p e rs o n c a n b e an aly tic an d au tocratic, o r co n su ltativ e (w ith individuals) a n d heuristic.

72 Part I • D ecisio n M aking and Analytics: An Overview

F o r a co m p u te riz e d sy stem to s u cce ssfu lly s u p p o rt a m a n a g e r, it sh o u ld fit th e d e c is io n situ atio n as w e ll as th e d e c is io n sty le. T h e re fo re , th e sy stem s h o u ld b e fle x ib le an d a d a p ta b le to d iffe ren t u sers. T h e ab ility to a s k w h a t-if an d g o a l-s e e k in g q u e s tio n s p ro v id e s fle x ib ility in this d irectio n . A W e b -b a s e d in te rfa c e u sin g g ra p h ics is a d esira b le fe a tu re in su p p o rtin g ce rta in d e c is io n styles. I f a D SS is to s u p p o rt v arying styles, skills, and k n o w le d g e , it sh o u ld n o t attem p t to e n fo r c e a s p e c ific p ro c e s s . R ather, it sh o u ld h e lp d e c is io n m a k e rs use a n d d e v e lo p th e ir o w n sty les, s k ills, an d k n o w le d g e.

D ifferen t d e c is io n sty les re q u ire d ifferen t typ es o f su p p ort. A m a jo r fa cto r th at d eter­ m in es th e ty p e o f su p p o rt requ ired is w h e th e r th e d e c is io n m a k e r is a n individual o r a group. Individual d e c is io n m akers n e e d a c c e s s to d ata and to e x p e rts w h o ca n p rovide a d v ice, w h e re a s gro u p s ad d itionally n e e d co lla b o ra tio n to ols. W e b -b a s e d D SS c a n p ro ­ v id e su p p o rt to b oth .

A lo t o f in form ation is a v ailab le o n th e W e b a b o u t co g n itiv e styles an d d ecisio n styles (e .g ., s e e B irk m an In tern atio n al, In c., birkm an.com ; K eirsey T e m p e ra m e n t Sorter and K e irse y T e m p e ra m e n t T h e o ry -II, keirsey.com ). M any p erson ality/ tem p eram en t tests are av ailab le to h e lp m an ag ers id entify th e ir o w n sty les an d th o s e o f th e ir em p lo y e es. Id entify ing a n individual’s style c a n h e lp e sta b lish th e m o st e ffe ctiv e co m m u n icatio n p attern s and id eal ta sk s fo r w h ich th e p e rs o n is suited.

D E C IS IO N M A K E R S D ecisio n s are o ften m ad e b y individuals, esp ecially a t low er m anage­ rial levels and in sm all organizations. T h e re m ay b e conflicting o b jectiv es e v e n fo r a sole d ecisio n m aker. F o r exam p le, w h e n m aking an investm ent decision, a n individual investor m ay co n sid er th e rate o f return o n th e investm ent, liquidity, and safety as objectives. Finally, d ecisio ns m ay b e fully autom ated (b u t only after a h u m an d ecisio n m aker d ecid es t o d o so!).

T h is d iscu ssio n o f d e cisio n m aking fo cu se s in large part o n a n individual d ecisio n m aker. M ost m ajo r d ecisio n s in m ed iu m -sized an d la rg e organ ization s are m ad e b y groups. O bv io u sly , th e re are o ften con flictin g o b je ctiv e s in a g ro u p d ecisio n -m ak in g setting. G roups c a n b e o f v ariab le size and m ay in clu d e p e o p le fro m d ifferent d ep artm ents o r fro m differ­ e n t org anizations. C ollaborating individuals m ay h av e d ifferent co g n itiv e styles, personality types, and d e cisio n styles. S o m e clash , w h ere a s o th ers are m utually e n h an cin g . C o nsensu s ca n b e a difficult p o litical p ro b lem . T h e re fo re , th e p ro c e s s o f d e c is io n m ak in g b y a group c a n b e very co m p licate d . C o m p u terized su p p ort c a n greatly e n h a n c e gro u p d ecisio n m aking. C o m p u ter su p p o rt can b e provid ed at a b ro a d lev el, e n a b lin g m e m b ers o f w h o le dep artm ents, divisions, o r e v e n entire organ ization s to co lla b o ra te on lin e. S u ch supp ort has e v o lv ed o v e r the p a st fe w y ears into en terp rise inform ation system s (E IS) and includes gro u p su p p ort system s (G S S ), e n terp rise re so u rce m a n a g e m e n t (ERM )/enterprise resou rce p lan n in g (E R P ), su p p ly c h a in m a n a g e m e n t (SC M ), k n o w le d g e m an ag e m e n t system s (KM S), and cu sto m e r relatio n sh ip m an ag e m e n t (CRM) system s.

SECTION 2 .2 REVIEW QUESTIONS

1 . W h a t are th e v arious a s p e cts o f d e cisio n m aking? 2 . Id en tify sim ilarities an d d iffe ren ce s b e tw e e n individ ual an d g ro u p d e cisio n m aking. 3 - D e fin e d ecisio n style an d d e s c rib e w h y it is im p o rtan t to c o n s id e r in th e d e c is io n ­

m ak in g p ro ce ss. 4 . W h a t are th e b e n e fits o f m ath em atical m odels?

2.3 PH A SES OF THE D ECISIO N -M A KIN G PR O CESS It is ad visable to fo llo w a system atic d ecisio n -m ak in g pro cess. Sim on (1 9 7 7 ) said that this involves th ree m ajo r p h ases: in tellig en ce, d esign, an d c h o ic e . H e later ad d ed a fourth p hase, im plem entation. M onitoring c a n b e co n sid ered a fifth p h ase— a fo rm o f fe e d b ack . H ow ever.

Chapter 2 • Foundations and T ech n o lo g ies for D ecisio n Making

S u c c e s s V

Organization objectives Search and scanning procedures

| Simplification Data collection Assumptions Problem identification

Problem ownership Problem classification Problem statement

Validation of the model

Verification, testing of proposed solution

Implementation of solution

Problem State m en t

Formulate a model Se t criteria for choice Search for alternatives Predict and measure outcomes

Solution to the model Sensitivity analysis Selection of the best [good]

alternative^) Plan for implementation

Failure

FIGURE 2.1 The Decisior-Making/Modeling Process.

w e v iew m onitoring a s th e intelligence p h a s e ap p lied to the im plem entation p h a se. Sim on 's m odel is th e m o st c o n c is e and y e t co m p le te characterization o f rational d ecisio n m aking. A con cep tu al picture o f th e d ecisio n -m akin g p ro cess is sh o w n in Figure 2.1.

T h e re is a c o n tin u o u s flo w o f activity fro m in te llig e n ce to d esig n to c h o ic e (s e e the b o ld lin es in Figure 2 .1 ), b u t a t a n y p h a s e , th e re m ay b e a return to a p rev iou s p h a se fe e d b a c k ). M odeling is a n e sse n tia l part o f this p ro ce ss. T h e see m in g ly c h a o tic n a tu re o f

fo llow in g a h a p h a z a rd p ath fro m p ro b le m d isco v e ry to so lu tio n v ia d e c is io n m ak in g can b e e x p la in e d b y th e s e fe e d b a c k lo o p s.

T h e d ecisio n -m akin g p ro cess starts w ith the intelligence phase; in this p h ase , the decision m aker ex a m in e s reality an d identifies and d efines th e problem . Problem ow nership s established as w ell. In th e design phase, a m o d el that represents th e system is constructed . This is d o n e b y m aking assum ptions that sim plify reality and w riting d ow n the relationships im o n g all the variables. T h e m o d el is th e n validated, an d criteria are determ ined in a princi­ ple o f c h o ic e fo r evaluation o f the alternative cou rses o f a ction that are identified. O ften, the process o f m o d el d ev elop m en t identifies alternative solutions an d v ice versa.

T h e ch o ice p h ase in clu d e s s e le c tio n o f a p ro p o s e d so lu tio n to th e m o d e l (n o t necessarily to th e p ro b le m it re p re sen ts). T h is so lu tio n is te sted to d eterm in e its viability. W h en th e p ro p o s e d so lu tio n se e m s re a s o n a b le , w e are re ad y fo r th e last p h a se : im p le­ m e n tation o f th e d e c is io n (n o t n e cessa rily o f a sy stem ). S u cc essfu l im p lem e n ta tio n results m solving th e re al p ro b le m . F ailu re lead s to a return to a n e a rlie r p h a s e o f th e p ro c e s s . In fact, w e c a n retu rn to a n e a rlie r p h a s e during an y o f th e latter th re e p h a se s. T h e d e c is io n ­ m aking situ ation s d e s c rib e d in th e o p e n in g v ig n e tte fo llo w S im o n ’s fo u r-p h a se m o d e l, as do alm o st all o th e r d e cisio n -m a k in g situations. W e b im p acts o n th e fo u r p h a se s, a n d vice versa, a re s h o w n in T a b le 2.1.

7 4 Part I • D ec isio n M aking and Analytics: An Overview

T A B L E 2.1 Sim on's Four Phases o f Decision M aking a n d th e W e b

Phase W e b Impacts Im pacts on th e W e b

Intelligence Access to information to identify problems and opportunities from internal and external data sources

Access to analytics methods to identify opportunities

Collaboration through group support systems (GSS) and knowledge management systems (KMS)

Identification of opportunities for e-commerce, W eb infrastructure, hardware and software tools, etc.

Intelligent agents, which reduce the burden of information overload

Smart search engines

Design Access to data, models, and solution methods

Use of online analytical processing (OLAP), data mining, and data warehouses

Collaboration through GSS and KMS Similar solutions available from KMS

Brainstorming methods (e.g., GSS) to collaborate in W eb infrastructure design

Models and solutions of W eb infrastructure issues

Choice Access to methods to evaluate the impacts of proposed solutions

Decision support system (DSS) tools, which examine and establish crite-a from models to determine Web, intranet, and extranet infrastructure

DSS tools, which determine how to route messages

Implementation Web-based collaboration tools (e.g., GSS) and KMS, which can assist in implementing decisions

Tools, which monitor the performance of e-commerce and other sites, including intranets, extranets, and the Internet

Decisions implemented on browse' and server design and access, which ultimately determined ho* j to set up the various com ponent " that have evolved into the Internet

N ote that th e re are m an y o th e r d ecisio n -m a k in g p ro c e s s e s. N otable am o n g t h e n 3 th e K e p n e r-T re g o e m e th o d (K e p n e r a n d T re g o e , 1 9 9 8 ), w h ich h a s b e e n a d o p te d b y man firm s b e c a u s e its to o ls a re read ily a v ailab le fro m K e p n e r-T re g o e , In c. (kepner-tregc»±. com). W e h av e fo u n d th at th e s e altern ativ e m o d e ls, in clu d in g th e K e p n e r-T re g o e m e ::.a read ily m a p in to S im o n ’s fo u r-p h a se m o d el.

W e n e x t tu rn to a d eta iled d iscu ssio n o f th e fo u r p h a s e s identified b y Sim on.

SECTION 2 . 3 REVIEW QUESTIONS

1 . List and b rie fly d e s c rib e S im o n ’s fo u r p h a s e s o f d e c is io n m aking.

2. W h a t are th e im p acts o f th e W e b 011 th e p h a se s o f d e cisio n m aking?

2.4 D ECISIO N M AKING: THE IN TELLIG EN C E PH ASE In te llig e n ce in d e c is io n m ak in g in v olv es sca n n in g th e e n v iro n m en t, e ith e r in te rm itte n t o r co n tin u o u sly . It in clu d e s sev era l activities a im ed a t id entifying p ro b le m situ a tio n ' 1 op p o rtu n ities. It m ay a ls o in clu d e m o n ito rin g th e results o f th e im p lem e n tatio n pha>r o a d ecisio n -m a k in g p ro cess.

Chapter 2 • Foundations and T ech n o lo g ies for D ecisio n Making 7 5

Problem (or O pportunity) Identification TEir intelligence p h a se b eg in s w ith th e identification o f organizational g oals and ob jectives e a c r d to an issue o f c o n c e rn (e .g ., inventory m anagem ent, jo b selectio n , lack o f o r in correct

W e r o re se n ce ) a n d determ ination o f w h eth er th ey are b ein g m et. P rob lem s o ccu r b e c a u se o f f a s a s f a c t i o n w ith th e status q u o. D issatisfaction is th e result o f a d ifferen ce b e tw e e n w hat

sien-cie desire (o r e x p e c t) an d w h at is occurring. In this first p h ase, a d ecisio n m ak e r attem pts i determine w h eth er a p ro b lem exists, identify its sym ptom s, d eterm ine its m agnitude, and

E m i l y d efine it. O ften, w h at is d escribed as a p ro b lem (e .g ., exce ssiv e co sts) m ay b e iv 2 sym ptom (i.e., m easu re) o f a p ro b lem (e .g ., im proper inventory levels). B e c a u s e real-

E ad d problem s are usu ally com plicated b y m an y interrelated factors, it is som etim es difficult D ifd n g u is h b e tw e e n th e sym ptom s and th e real problem . N ew opportunities an d prob-

t - s certainly m ay b e u n co v e re d w h ile investigating th e cau ses o f sym ptom s. F or exam p le, pciieation C ase 2.1 d escrib e s a classic story o f recognizin g th e correct problem .

T h e e x is te n c e o f a p ro b le m ca n b e d eterm in ed b y m o n ito rin g an d an aly zin g th e e s s m iz a tio n ’s p ro d u ctivity lev el. T h e m e a su re m e n t o f produ ctivity a n d th e co n stru ctio n r f a m o d el are b a se d o n re al d ata. T h e c o lle c tio n o f data an d th e estim atio n o f fu ture data

s - e am o n g th e m o st d ifficu lt s te p s in th e analysis. T h e fo llo w in g are s o m e issu es th a t m ay d uring d ata c o lle c tio n an d estim atio n an d thu s p lag u e d e cisio n m akers:

• D ata are n o t a v ailab le. As a result, th e m o d el is m ad e w ith, a n d re lie s o n , p o ten tially

in accu rate estim ates. • O b tain in g d ata m ay b e e x p e n siv e . • D ata m ay n o t b e a ccu ra te o r p re c ise e n o u g h . • D ata e stim a tio n is o fte n su b jectiv e. • D ata m ay b e in secu re. • Im p o rtan t data th at in flu e n ce th e results m ay b e qualitative (so ft). • T h e re m ay b e to o m an y d ata (i.e ., in form ation o v erlo ad ).

Application Case 2.1 Making Elevators Go Faster! This story h a s b e e n re p o rted in n u m e ro u s p la ce s and h a s alm o st b e c o m e a c la ss ic e x a m p le to e x p la in d ie n e e d fo r p ro b le m id entification. A c k o ff (a s cited in L arson, 1 9 8 7 ) d e s c rib e d th e p ro b le m o f m an agin g com p laints a b o u t slo w elev ato rs in a tall h otel tow er. After trying m a n y so lu tio n s fo r re d u cin g th e c o m ­ plaint: stag g erin g e le v ato rs to g o to d ifferen t floors, adding o p erato rs, a n d s o o n , th e m a n a g e m e n t d e te r­ m ined that th e re a l p ro b le m w a s n o t a b o u t th e a c tu a l w aiting tim e b u t rath er th e p e r c e iv e d w aitin g tim e. So th e so lu tio n w a s to install fu ll-len gth m irrors o n e le v a to r d o o rs o n e a c h flo o r. As H e s se a n d W o o ls e y (1 9 7 5 ) p u t it, “th e w o m e n w o u ld lo o k a t th e m se lv es in th e m irrors a n d m a k e ad ju stm ents, w h ile th e m e n w ould lo o k a t th e w o m e n , a n d b e fo r e th ey k n e w it, th e e le v a to r w as th e r e .” B y red u cin g th e p e rce iv ed w aiting tim e, th e p ro b le m w e n t aw ay. B a k e r and

C am eron ( 1 9 9 6 ) give sev eral o th e r e x a m p le s o f d is­ tractio n s, in clu d in g lig h tin g, d isplays, an d s o o n , that org an izatio n s u s e to re d u c e p e rc e iv e d w aiting tim e. I f th e real p ro b le m is id e n tified a s p e r c e iv e d w aiting tim e, it c a n m a k e a b ig d iffe re n ce in th e p ro p o s e d so lu tio n s an d th eir c o s ts . F o r e x a m p le , fu ll-len gth m irrors p ro b a b ly c o s t a w h o le lo t less th a n ad ding a n elevator!

Sources: Based on J. Baker and M. Cameron, “The Effects of the Service Environment on Affect and Consumer Perception of Waiting Time: An Integrative Review and Research Propositions,” Jo u rn a l o f th e A cadem y o f M arketin g S cience, Vol. 24, September 1996, pp. 338-349; R. Hesse and G . Woolsey, A pplied M anagem ent Scien ce: A Q uick a n d Dirty A pproach, SRA Inc., Chicago, 1975; R. C. Larson. “Perspectives on Queues: Social Justice and the Psychology of Queuing,” O peration s R esearch, Vol. 35, No. 6, November/December 1987, pp. 895-905.

• O u tc o m e s ( o r re s u lts ) m a y o c c u r o v e r a n e x te n d e d p e rio d . As a re su lt, re v ­ e n u e s , e x p e n s e s , a n d p ro fits w ill b e r e c o r d e d a t d iffe re n t p o in ts in tim e . T o o v e r c o m e th is d ifficu lty , a p re s e n t-v a lu e a p p r o a c h c a n b e u s e d i f th e re su lts are

q u a n tifia b le . • It is a ssu m e d th at future d ata w ill b e sim ilar to h istorical data. I f this is n o t th e ca se ,

th e natu re o f th e c h a n g e h a s to b e p re d icte d a n d in clu d e d in th e analysis.

W h e n th e p relim in ary in v estigation is co m p le te d , it is p o ss ib le to d eterm in e w h e th e r a p ro b le m really exists, w h e re it is lo ca te d , an d h o w sig n ifican t it is. A k e y issu e is w h e th e r a n in fo rm atio n sy stem is rep ortin g a p ro b le m o r o n ly th e sym ptom s o f a p ro b lem . For ex a m p le , if rep o rts in d icate th at s a le s are d o w n , th e re is a p ro b lem , b u t th e situation, n o d o u b t, is’ sy m p to m atic o f th e p ro b lem . It is critical to k n o w th e real p ro b lem . Som etim es it m ay b e a p ro b le m o f p e rce p tio n , in cen tiv e m ism atch , o r organ ization al p ro c e s s e s rath er

th a n a p o o r d e c is io n m odel.

Problem Classification P ro b le m classificatio n is th e co n c e p tu a liz a tio n o f a p ro b lem in a n attem p t to p la c e it in a d e fin a b le categ ory , p o ssib ly lead in g to a stan d ard so lu tio n ap p ro a ch . An im portant a p p ro a ch classifies p ro b lem s a cco rd in g to th e d e g r e e o f stru ctu red n ess e v id e n t in them . T h is ran g es fro m totally stru ctured ( i.e ., p ro g ra m m e d ) to to tally u n sta ictu re d ( i.e ., u n p ro­ gram m ed ), as d e s crib e d in C h ap ter 1.

Problem Decomposition M any c o m p le x p ro b le m s c a n b e divided into su b p ro b le m s. So lv in g th e sim p ler su b p ro b ­ lem s m ay h e lp in solv in g a c o m p le x p ro b lem . A lso, see m in g ly p o o rly stru ctured p ro b lem s s o m e tim e s h a v e h ighly stru ctured s u b p ro b lem s. J u s t as a sem istru ctu red p ro b le m results w h e n s o m e p h a se s o f d e c is io n m ak in g are stru ctu red w h e re a s o th e r p h a se s are un struc­ tured, so w h e n so m e s u b p ro b lem s o f a d ecisio n -m a k in g p ro b le m are stru ctured w ith o th ers un structured , th e p ro b le m its e lf is sem istru ctu red . As a D SS is d e v e lo p e d an d th e d e c is io n m a k e r an d d ev e lo p m e n t sta ff learn m o re a b o u t th e p ro b le m , it gains structure. D e c o m p o sitio n a lso facilitates co m m u n ica tio n a m o n g d e c is io n m akers. D e co m p o sitio n is o n e o f th e m o st im portant a s p e cts o f th e an aly tical h iera rch y p ro ce ss. (AHP is d iscu sse d in C h ap ter 1 1, w h ich h e lp s d e c is io n m a k e rs in co rp o ra te b o th qualitative an d quantitative fa cto rs into th eir d ecisio n -m ak in g m o d e ls.)

Problem Ownership In th e in te llig e n c e p h a s e , it is im p o rta n t to e s ta b lis h p r o b le m o w n e rs h ip . A p ro b le m e x is ts in a n o rg a n iz a tio n o n ly if s o m e o n e o r s o m e g ro u p ta k e s o n th e re s p o n s ib ility o a tta c k in g it a n d i f t h e o r g a n iz a tio n h a s th e ab ility to s o lv e it. T h e a s s ig n m e n t o f a u th o r­ ity to s o lv e th e p r o b le m is c a lle d p ro b lem ow n ersh ip . F o r e x a m p le , a m a n a g e r m ay f e e l th a t h e o r s h e h a s a p r o b le m b e c a u s e in te r e s t ra te s a re to o h ig h . B e c a u s e in te re st ra te le v e ls a r e d e te rm in e d at th e n a tio n a l a n d in te rn a tio n a l le v e ls , a n d m o st m a n a g e rs c a n d o n o th in g a b o u t th e m , h ig h in te re s t rate s a r e th e p r o b le m o f th e g o v e r n m e n t, n o t a p r o b le m fo r a s p e c ific c o m p a n y to s o lv e . T h e p r o b le m c o m p a n ie s a c tu a lly fa c e is h o w to o p e r a te in a h ig h -in te r e s t-r a te e n v iro n m e n t. F o r a n in d iv id u al c o m p a n y , th e in te re s t ra te le v e l sh o u ld b e h a n d le d as a n u n c o n tr o lla b le (e n v ir o n m e n ta l) fa c to r to b e

p re d ic te d . , . W h e n p ro b le m o w n e rsh ip is n o t e sta b lish e d , e ith e r s o m e o n e is n o t d o in g his or

h e r jo b o r th e p ro b le m at h an d h a s y e t to b e id e n tified as b e lo n g in g to a n y o n e. It is th e n im portant fo r s o m e o n e to e ith e r v o lu n te e r to o w n it o r assig n it to so m e o n e .

T h e in te llig e n ce p h a s e e n d s w ith a fo rm al p ro b le m statem en t.

7 6 Part I • D ec isio n M aking and Analytics: An O verview

Chapter 2 * Foundations and T ech n o lo g ies for D ecisio n M aking 7 7

f ZCTION 2 . 4 R EV IEW QUESTIONS

1 . W hat is th e d iffe re n ce b e tw e e n a p ro b le m a n d its sym ptom s?

2 . W h y is it im p o rtan t to classify a problem ? 3 . W hat is m e a n t b y p r o b le m decom p osition ? -L W hy is e sta b lish in g p ro b le m o w n e rsh ip s o im portant in th e d e cis io n -m a k in g process?

2.5 DECISIO N M AKIN G : THE D ESIG N PH ASE The d esig n p h a s e in v olv es find ing o r d ev elo p in g an d a n aly zin g p o ss ib le co u rs e s o f a ctio n These in clu d e u n d erstan d in g th e p ro b le m a n d testin g so lu tio n s fo r fe a sib ility A m o d el rr th e d e cisio n -m a k in g p ro b le m is co n stru cte d , te sted , an d validated . Let u s first d efin e

i m odel.

Models1 A m ajor characteristic o f a D SS and m any B l to o ls (n o ta b ly th o se o f b u sin ess an aly tics) is the inclu sion o f a t le a st o n e m o d el. T h e b a sic idea is to p erfo rm th e D SS analysis o n a m o d el o f reality rath er th an o n the real system . A m o d el is a sim plified rep resen tatio n o r ab strac­ tio n o f reality. It is usually sim plified b e ca u se reality is to o co m p le x to d escrib e e xactly an d b ecau se m u ch o f th e com p lex ity is actu ally irrelevant in solving a sp e cific p ro blem .

Mathem atical (Quantitative) Models T h e com p lexity o f relationships in m an y organizational system s is d escrib e d m athem ati­ cally. M ost D SS an alyses are p erfo rm ed num erically w ith m ath em atical o r o th e r quantitative

m odels.

The Benefits of Models We u se m o d els fo r th e fo llo w in g re aso n s:

• M anipu lating a m o d e l (ch a n g in g d e c is io n v ariab les o r th e e n v iro n m e n t) is m u ch e a sie r th a n m an ip u latin g a real system . E x p e rim e n tatio n is e a s ie r an d d o e s n o t in terfere w ith th e o rg an izatio n ’s daily o p eratio n s.

• M odels e n a b le th e c o m p re ss io n o f tim e. Y e a r s o f o p era tio n s c a n b e sim u lated in

m in u tes o r s e c o n d s o f co m p u te r tim e. • T h e c o s t o f m o d e lin g analysis is m u ch lo w e r th a n th e c o s t o f a sim ilar e x p e rim e n t

c o n d u cte d o n a re al system . • T h e c o s t o f m ak in g m istak es during a trial-an d -error e x p e rim e n t is m u ch lo w e r

w h e n m o d e ls are u s e d th a n w ith real system s. • T h e b u s in e s s e n v iro n m en t in v olv es co n s id e ra b le u n certainty. W ith m o d elin g , a

m an ag e r c a n estimate th e risks resultin g fro m s p e c ific actions. • M ath em atical m o d e ls e n a b le th e an alysis o f a very large, s o m e tim e s in fin ite, n u m b e r

o f p o ss ib le solu tion s. E v e n in s im p le p ro b le m s, m an ag e rs o fte n h av e a larg e n u m b e r

o f a ltern ativ es fro m w h ich to c h o o s e . • M o d els e n h a n c e an d re in fo rce learn in g and training. • M o d els a n d so lu tio n m e th o d s are read ily available.

M o d e lin g in v o lv e s co n c e p tu a liz in g a p ro b le m an d a b stra ctin g it t o q u an titativ e and/or q u a lita tiv e fo rm ( s e e C h ap te r 9 ) . F o r a m a th em a tica l m o d e l, th e v a r ia b le s are

-Caution- Many students and professionals view models strictly as those of “data modeling’ in the context of S te r n s a n a " ! and design. Here, we consider analytical models such as those of linear programmmg, s t a ­ tion, and forecasting.

7 8 P art I • D e c is io n M aking and Analytics-. An O verview

id e n tified , a n d th e ir m u tu al re la tio n sh ip s a re e sta b lis h e d . S im p lifica tio n s a re m ad e , w h e n e v e r n e c e s s a ry , th ro u g h a ssu m p tio n s. F o r e x a m p le , a re la tio n sh ip b e tw e e n tw o v a r ia b le s m ay b e a ssu m e d to b e lin e a r e v e n th o u g h in re a lity th e r e m ay b e s o m e n o n ­ lin e a r e ffe c ts . A p r o p e r b a la n c e b e tw e e n th e le v e l o f m o d e l s im p lifica tio n a n d th e re p ­ r e s e n ta tio n o f re ality m u st b e o b ta in e d b e c a u s e o f th e c o s t—b e n e fit tra d e -o ff. A s im p le r m o d e l lead s to lo w e r d e v e lo p m e n t c o s ts , e a s ie r m a n ip u la tio n , a n d a fa s te r s o lu tio n b u t is le s s re p re s e n ta tiv e o f th e re a l p ro b le m a n d c a n p r o d u c e in a c c u ra te resu lts. H o w e v e r, a sim p le r m o d e l g e n e ra lly re q u ire s fe w e r d ata, o r th e d ata a re a g g re g a te d an d e a s ie r

to o b ta in . T h e p ro c e s s o f m o d e lin g is a co m b in a tio n o f art a n d s c ie n c e . As a s c ie n c e , th e re

a re m an y stan d ard m o d el cla ss e s av a ila b le, and , w ith p ra ctice , a n an aly st c a n d e te im in e w h ic h o n e is a p p lic a b le to a g iv e n situ ation . As an art, creativity an d fin e s s e a re requ ired w h e n d eterm in in g w h at sim p lifying a ssu m p tio n s c a n w o rk , h o w to c o m b in e ap p ro p ri­ a te fe a tu re s o f th e m o d e l cla ss e s , an d h o w to in te g rate m o d e ls to o b ta in valid solu tion s. M o d e ls h av e decision variables that d e s c rib e th e altern ativ es fro m a m o n g w h ic h a m a n a g e r m u st c h o o s e (e .g ., h o w m an y ca rs to d e liv e r to a s p e c ific re n tal ag e n cy , h o w to ad vertise at s p e c ific tim e s, w h ic h W e b serv e r to b u y o r le a s e ), a resu lt v a ria b le o r a s e t o f resu lt v a ria b le s ( e .g ., p rofit, re v e n u e , s a le s ) th a t d e s c r ib e s th e o b je c tiv e o r g o a l o f th e d e cis io n -m a k in g p ro b le m , and u n c o n tro lla b le v a ria b le s o r p a ra m ete rs (e .g ., e c o n o m ic c o n d itio n s ) th a t d e s c rib e th e e n v iro n m en t. T h e p r o c e s s o f m o d e lin g in v o lv e s d eterm in ­ ing th e (u su ally m ath em atical, s o m e tim e s s y m b o lic ) re la tio n sh ip s am o n g th e v ariab le s. T h e s e to p ic s a re d iscu sse d in C h ap ter 9-

Selection o f a Principle of Choice A p rinciple o f ch o ice is a c rite rio n th at d e s c r ib e s th e a c c e p ta b ility o f a so lu tio n a p p ro a c h . In a m o d e l, it is a re su lt v aria b le . S e le c tin g a p rin cip le o f c h o ic e is n o t part o f th e c h o ic e p h a s e b u t in v o lv e s h o w a p e rs o n e sta b lis h e s d e cis io n -m a k in g o b je c tiv e (s ) an d in c o rp o ra te s th e o b je c tiv e (s ) in to th e m o d e l(s ). A re w e w illin g to a ssu m e high risk, o r d o w e p re fe r a lo w -risk a p p ro ach ? A re w e a tte m p tin g to o p tim ize o r satisfice? It is a ls o im p o rtan t to r e c o g n iz e th e d iffe re n ce b e tw e e n a crite rio n a n d a co n strain t ( s e e T e c h n o lo g y In sig h ts 2 .1 ). A m o n g th e m an y p rin cip le s o f c h o ic e , n o rm ativ e and d escrip tiv e a re o f p rim e im p o rtan ce .

T E C H N O L O G Y IN SIG H T S 2 . 1 T h e D if f e r e n c e B e t w e e n a C r ite r io n a n d a C o n s tr a in t

Many people new to the formal study o f decision making inadvertently confuse the concepts o f criterion and constraint. Often, this is because a criterion may imply a constraint, either implicit or explicit, thereby adding to the confusion. For example, there may b e a distance criterion that the decision maker does not want to travel too far from home. However, there is an implicit constraint that the alternatives from which he selects must be within a certain distance from his home. This constraint effectively says that if the distance from home is greater than a certain amount, then the alternative is not feasible— or, rather, the distance to an alternative must be less than or equal to a certain number (this would b e a formal relationship in some models; in the model in this case, it reduces the search, considering fewer alternatives). This is similar to what happens in some cases when selecting a university, where schools beyond a single d ays diiv- ing distance would not be considered by most people, and, in fact, the utility function (criterion value) o f distance can start out low close to home, peak at about 70 miles (about 100 km)— say, the distance betw een Atlanta (home) and Athens, Georgia— and sharply drop off thereafter.

C hapter 2 • Foundations and T ech n o lo g ies for D ecisio n Making 7 9

Normative Models Normative m odels a re m o d e ls in w h ich th e c h o s e n alternative is d em o n stra b ly th e b e s t o f all p o ssib le altern ativ es. T o find it, th e d e c is io n m a k e r sh o u ld e x a m in e all th e altern a­ t e s a n d p ro v e th a t th e o n e s e le c te d is in d e e d th e b e s t, w h ic h is w h a t th e p e rs o n w o u ld no n n ally w an t. T h is p ro c e s s is b a sica lly optim ization. T h is is ty p ically th e g o a l o f w h at w e call p rescrip tiv e analytics (P a rt IV ). In o p era tio n a l term s, o p tim ization c a n b e a ch iev ed

in o n e o f th re e w ays:

1 . G e t th e h ig h est lev el o f g o al attain m en t fro m a g iv e n s e t o f re s o u rce s . F o r e x a m p le , w h ich altern ativ e w ill y ield th e m ax im u m p ro fit fro m an in v e stm e n t o f $ 1 0 m illion?

2 . Fin d th e altern ativ e w ith th e h ig h est ratio o f g o al attain m en t to c o s t (e .g ., profit p e r d ollar in v e ste d ) o r m ax im ize productivity.

3 . Fin d th e altern ativ e w ith th e lo w e st c o s t (o r sm allest am o u n t o f o th e r r e s o u r c e s ) that w ill m e e t a n a c c e p ta b le lev el o f goals. F o r e x a m p le , if y o u r ta sk is to s e le c t hard w are fo r a n in tran et w ith a m inim um b an d w id th , w h ic h altern ativ e w ill a c c o m p lis h this g o al a t th e le a st cost?

N orm ative d e c is io n th e o ry is b a s e d o n th e fo llo w in g assu m p tio n s o f rational

d ecision m akers:

• H um ans are e c o n o m ic b e in g s w h o s e o b je c tiv e is to m ax im ize th e a tta in m e n t o f goals; th a t is, th e d e c is io n m ak er is rational. (M ore o f a g o o d th in g [re v en u e, fun] is b etter th a n le ss; le s s o f a b a d th in g [cost, pain] is b e tte r than m o re .)

• F o r a d e cisio n -m a k in g situation, all v ia b le alternative co u rs e s o f a ctio n an d th eir c o n s e q u e n c e s , o r at le a st th e p ro b ab ility a n d th e v a lu e s o f th e c o n s e q u e n c e s , are

k now n. • D e c is io n m a k e rs h av e a n o rd e r o r p re fe re n c e that e n a b le s th e m to ran k th e d esir­

ability o f all c o n s e q u e n c e s o f th e analysis (b e s t to w o rst).

Are d ecisio n m akers really rational? T h o u g h th ere m ay b e m ajor anom alies in th e pre­ sumed rationality o f financial an d e c o n o m ic behavior, w e tak e th e view that th e y cou ld b e caused b y in co m p e te n ce , lack o f kn ow led g e, multiple g oals b ein g fram ed inadequately, m is­ understanding o f a d ecisio n m aker’s true e x p e cte d utility, an d tim e-pressure im pacts. T h ere axe o th er anom alies, o ften cau sed b y tim e pressure. F o r exam p le, Stew art (2 0 0 2 ) described i. nu m ber o f research ers w orking w ith intuitive d ecisio n m aking. I h e idea o f thinking with vour gut” is obviou sly a heuristic ap p ro ach to d ecisio n making. It w orks w ell fo r firefighters m d military person n el o n th e battlefield. O n e critical asp ect o f d ecisio n m aking in this m ode is that m any scen ario s hav e b e e n thought through in ad vance. Even w h e n a situation is new . i c a n quickly b e m atch ed to a n existing o n e on-the-fly, an d a re aso n ab le solu tion can b e obtained (through p attern recognition). Luce e t al. (2 0 0 4 ) d escribed h o w em o tio n s affect d ecision m aking, an d Pauly (2 0 0 4 ) d iscu ssed in con sisten cies in d ecisio n m aking.

W e b e lie v e th at irrationality is ca u se d b y th e facto rs liste d p reviou sly. F o r e x a m ­ p le, T v ersk y e t al. (1 9 9 0 ) investigated th e p h e n o m e n o n o f p re fe re n c e rev ersal, w h ich is a k n o w n p ro b le m in ap p ly in g th e AHP to p ro b lem s. A lso, s o m e criterio n o r p re fe re n c e m ay b e om itted fro m th e analy sis. R atn er e t al. (1 9 9 9 ) in vestigated h o w v arie ty c a n ca u se individuals to c h o o s e less-p re fe rre d o p tio n s, e v e n th o u g h th e y will e n jo y th e m less. B u t w e m ain tain th at v ariety cle a rly has valu e, is p art o f a d e c is io n m a k e r’s utility, a n d is a criterion and/or co n stra in t th at sh o u ld b e c o n s id e re d in d e c is io n m aking.

Suboptimization B y d efin ition , op tim izatio n re q u ires a d e c is io n m a k e r to co n s id e r th e im p act o f e a c h alter­ native c o u rs e o f a c tio n o n th e e n tire org an izatio n b e c a u s e a d e c is io n m ad e in o n e area m ay h av e sig n ifican t e ffe c ts (p o sitiv e o r n e g a tiv e ) o n o th e r areas. C o nsid er, fo r e x a m p le , a

8 0 Part I • D ec isio n M aking and Analytics: An Overview

m ark etin g d ep artm en t that im p lem e n ts a n e le c tro n ic c o m m e rc e (e -c o m m e r c e ) site. W ithin hou rs, o rd ers far e x c e e d p ro d u ctio n cap acity . T h e p ro d u ctio n d ep artm en t, w h ic h plans its o w n sch e d u le , ca n n o t m e e t d em an d . It m ay g e a r up fo r a s high d em an d as p o ssi­ b le . Id eally a n d in d e p e n d en tly , th e d ep artm en t sh o u ld p ro d u ce o n ly a fe w p ro d u cts in ex tre m e ly larg e q u an tities to m inim ize m an u factu rin g co sts. H ow ever, s u c h a p la n m ight result in large, co stly in v en to ries and m ark etin g d ifficu lties ca u se d b y th e la c k o f a variety o f p ro d u cts, e sp e c ia lly i f c u sto m ers start to c a n c e l o rd ers th a t are n o t m e t in a tim ely w ay. T h is situ ation illustrates th e s eq u en tial n atu re o f d e c is io n m aking.

A sy stem s p o in t o f v ie w a ss e s s e s th e im p act o f e v ery d e c is io n o n th e en tire sys­ tem . T h u s, th e m ark etin g d ep artm en t sh o u ld m a k e its plans in co n ju n ctio n w ith o th er d ep artm en ts. H o w ev er, s u ch a n a p p ro a ch m ay re q u ire a co m p lica te d , e x p e n s iv e , tim e- c o n su m in g analysis. In p ra ctice , th e MSS b u ild e r m a y c lo s e th e sy stem w ithin n arrow b o u n d arie s, co n sid erin g o n ly th e part o f th e o rg an izatio n u n d e r study (th e m ark etin g and/ o r p ro d u ctio n d ep artm ent, in this c a s e ). B y sim plifying, th e m o d e l th e n d o e s n o t in co rp o ­ rate ce rta in co m p lica te d relatio n sh ip s th a t d e s c rib e in te ra ctio n s w ith an d am o n g th e o th er d ep artm en ts. T h e o th e r d ep artm en ts c a n b e a g g re g a te d in to sim p le m o d e l co m p o n e n ts. S u c h a n a p p ro a ch is ca lle d suboptim ization.

I f a su b o p tim al d e c is io n is m ad e in o n e p art o f th e org an izatio n w ith o u t co n sid erin g th e d etails o f th e re st o f th e o rg an ization , th e n a n o p tim al so lu tio n fro m th e p o in t o f v iew o f th a t part m ay b e in ferio r fo r th e w h o le . H o w e v e r, su b o p tim izatio n m ay still b e a very p ractical a p p ro a ch to d e cisio n m akin g, a n d m an y p ro b le m s are first a p p ro a c h e d fro m this p e rsp e ctiv e. It is p o ss ib le to re a c h ten tative c o n c lu sio n s (a n d g e n erally u s a b le resu lts) b y an aly zin g o n ly a p o rtio n o f a system , w ith o u t g e ttin g b o g g e d d o w n in to o m an y details. A fter a so lu tio n is p ro p o se d , its p o ten tia l e ffe c ts o n th e rem ain in g d ep artm en ts o f th e o rg an izatio n c a n b e tested . I f n o sig n ifican t n e g a tiv e e ffe cts are fo u n d , th e so lu tio n ca n b e im p lem en ted .

Su b op tim izatio n m ay a lso ap p ly w h e n sim p lifying assu m p tio n s are u s e d in m o d ­ e lin g a s p e cific p ro b lem . T h e re m ay b e to o m an y d etails o r to o m an y d ata to in co rp o rate in to a s p e c ific d ecisio n -m a k in g situation, an d s o n o t all o f th e m are u s e d in th e m o d el. I f th e so lu tio n to th e m o d e l s e e m s re a s o n a b le , it m ay b e v alid fo r th e p ro b le m a n d thus b e ad o p ted . F o r e x a m p le , in a p ro d u ctio n d ep artm e n t, parts are o fte n p artitio n ed into A/B/C inventory cate g o rie s. G e n e rally , A item s (e .g ., larg e g e ars, w h o le a s s e m b lie s ) are e x p e n s iv e (say , $ 3 ,0 0 0 o r m o re e a c h ), built to o rd e r in sm all b a tc h e s , an d in ven toried in lo w q u an tities; C item s (e .g ., nuts, b o lts, scre w s) a re very in e x p e n s iv e (say , less th a n $2) a n d o rd e re d a n d u se d in very larg e q u antities; an d B item s fall in b e tw e e n . All A item s ca n b e h a n d led b y a d etailed s ch e d u lin g m o d e l an d p h ysically m o n ito re d clo s e ly b y m an­ ag e m e n t; B item s are g e n erally so m e w h a t a g g reg ated , th eir g ro u p in g s are sch e d u le d , and m a n a g e m e n t rev iew s th e s e parts less freq u en tly ; a n d C item s are n o t sch e d u le d b u t are sim p ly a cq u ire d o r built b a se d o n a p o licy d e fin e d b y m a n a g e m e n t w ith a sim p le e c o ­ n o m ic ord e r qu antity (E O Q ) o rd erin g sy stem that a ssu m e s co n s ta n t an n u al d em an d . T h e p o licy m ig h t b e re v iew e d o n c e a year. T h is situ atio n a p p lie s w h e n d eterm in in g all criteria o r m o d e lin g th e en tire p ro b le m b e c o m e s p ro h ib itiv ely tim e -co n su m in g o r e x p e n s iv e .

Su b op tim izatio n m ay a ls o in v o lv e sim p ly b o u n d in g th e s e a rch fo r a n op tim u m (e .g ., b y a h e u ristic) b y co n sid e rin g fe w e r criteria o r alternativ es o r b y elim in atin g large p o rtio n s o f th e p ro b le m fro m evalu ation. I f it tak es to o lo n g to s o lv e a p ro b le m , a g o o d - e n o u g h so lu tio n fo u n d alread y m ay b e u s e d a n d th e o p tim ization e ffo rt term inated.

Descriptive Models D escriptive m odels d e s c rib e th in gs as th e y a re o r as th e y are b e lie v e d to b e . T h e s e m o d e ls are typ ically m ath em atically b a se d . D escrip tiv e m o d e ls are e x tre m e ly u sefu l in D SS fo r in vestigatin g th e c o n s e q u e n c e s o f vario u s alternative co u rs e s o f a c tio n u n d er

Chapter 2 • Foundations and T ech n o lo g ies for D ecisio n Making 81

d ifferent co n fig u ra tio n s o f inputs and p ro ce s s e s. H o w ev er, b e c a u s e a d e scrip tiv e analysis ch e c k s th e p e rfo rm a n c e o f th e sy stem fo r a g iv e n s e t o f alternativ es (ra th er th an fo r all alternatives), th e re is n o g u aran te e th a t a n altern ativ e s e le c te d w ith th e aid o f d escrip tive analysis is o p tim al. In m any c a s e s , it is o n ly satisfactory.

Sim ulation is p ro b a b ly t h e m o st co m m o n d escrip tiv e m o d e lin g m eth o d . S im u latio n is th e im itatio n o f reality a n d h a s b e e n a p p lie d to m an y a re a s o f d e c is io n m aking. C o m p u ter an d v id e o g a m e s a re a fo rm o f sim u lation: A n artificial reality is cre a te d , and th e g am e p la y er liv es w ith in it. Virtual reality is a lso a fo rm o f sim u latio n b e c a u s e th e en vi­ ro n m en t is sim u lated , n o t real. A co m m o n u s e o f sim u latio n is in m an u factu rin g. Again, co n sid er th e p ro d u ctio n d ep artm e n t o f a firm w ith co m p lica tio n s ca u se d b y th e m arketin g d ep artm ent. T h e ch aracteristics o f e a c h m a ch in e in a jo b s h o p a lo n g th e su p p ly ch a in c a n b e d e s c rib e d m ath em atically. R elation sh ip s c a n b e e sta b lis h e d b a s e d o n h o w e a c h m a ch in e p h y sically runs and re la tes to o th ers. G iv e n a trial s c h e d u le o f b a tc h e s o f parts, ir is p o ss ib le to m e a su re h o w b a tc h e s flo w th ro u g h th e system and to u s e th e statistics from e a c h m a ch in e . A lternative sch e d u le s m ay th e n b e tried a n d th e statistics re co rd e d until a re a s o n a b le s ch e d u le is fo u n d . M arketing c a n e x a m in e a c c e s s a n d p u rch a s e pat­ terns o n its W e b site. S im u lation c a n b e u se d to d eterm in e h o w to stru cture a W e b site fo r im proved p e rfo rm a n ce an d to e stim ate future p u rch ases. B o th d ep artm en ts c a n th e re fo re u se prim arily e x p e rim e n ta l m o d e lin g m ethod s.

C lasses o f d escrip tiv e m o d els in clu d e th e fo llow ing:

• C o m p lex in v e n to ry d ecisio n s • E n v iro n m en tal im p act analysis • Fin an cial p lan n in g • In fo rm atio n flow • M arkov an alysis (p red ictio n s) • S ce n a rio analysis • Sim ulation (altern ativ e typ es) • T e c h n o lo g ic a l fo recastin g • W aitin g -lin e (q u e u in g ) m a n a g e m e n t

A n u m b e r o f n o n m ath e m atical d escrip tiv e m o d e ls are av ailab le fo r d e c is io n m ak ­ ing. O n e is th e co g n itiv e m ap (s e e E d en a n d A ck erm an n , 2 0 0 2 ; an d Je n k in s , 2 0 0 2 ). A co g n itiv e m ap c a n h e lp a d e c is io n m a k e r sk e tc h o u t th e im portant q u alitativ e facto rs an d ih eir cau sal re la tio n sh ip s in a m essy d ecisio n -m a k in g situation. T h is h e lp s th e d e cisio n m aker (o r d ecisio n -m a k in g g ro u p ) fo c u s o n w h at is re le v a n t an d w h a t is not, and th e m ap e v o lv e s as m o re is learn ed a b o u t th e p ro b lem . T h e m ap c a n h e lp th e d e c is io n m ak er understand issu es b etter, fo c u s b ette r, and re a c h clo su re . O n e in terestin g so ftw a re to o l fo r co g n itiv e m a p p in g is D e c is io n E x p lo rer fro m B a n x ia Softw are Ltd. (b anxia.com ; try th e d em o ).

A n o th er d escrip tiv e d ecisio n -m ak in g m o d e l is th e u s e o f narratives t o d e s c rib e a d ecisio n -m ak in g situation. A n arrativ e is a sto ry th a t h e lp s a d e c is io n m a k e r u n c o v e r th e im portant a s p e cts o f th e situ ation a n d lead s to b e tte r u n d erstan d in g a n d fram ing . T h is is extrem ely e ffe c tiv e w h e n a g ro u p is m ak in g a d e cis io n , an d it c a n lea d to a m o re c o m ­ m on v ie w p o in t, a lso c a lle d a fr a m e . Ju rie s in co u rt trials typ ically u s e n arrativ e-b ased a p p ro ach e s in re a ch in g verd icts ( s e e A llan, Fram e, an d T u rn ey , 2 0 0 3 ; B e a c h , 2 0 0 5 ; an d D en n in g , 2 0 0 0 ).

Good Enough, or Satisficing A ccord ing to S im o n ( 1 9 7 7 ) , m o st h u m an d e c is io n m ak in g , w h e th e r o rgan izatio n al o r indi­ vidual, in v o lv es a w illin g n ess to settle fo r a satisfacto ry so lu tio n , “so m e th in g less th a n th e b e s t.” W h e n satisficing, th e d e c is io n m a k e r sets up a n asp iration, a g o a l, o r a d esired

8 2 Part I • D ec isio n M aking and Analytics: An Overview

lev el o f p e rfo rm a n ce a n d th e n s e a r c h e s th e altern ativ es un til o n e is fo u n d th at a ch iev es this lev el. T h e u su al re a s o n s fo r satisficin g are tim e p re ssu re s (e .g ., d ecisio n s m ay lo se v a lu e o v e r tim e ), th e ability to a ch ie v e o p tim ization (e .g ., solv in g s o m e m o d els cou ld ta k e a really lo n g tim e, a n d re co g n itio n th a t th e m argin al b e n e fit o f a b e tte r so lu tio n is n o t w o rth th e m argin al c o s t to o b tain it (e .g ., in s e a r c h in g th e In tern et, y o u ca n lo o k at on ly s o m any W e b sites b e fo re y o u ru n o u t o f tim e a n d e n e rg y ). In s u c h a situation, the d e c is io n m a k e r is b eh a v in g rationally, th o u g h in re ality h e o r s h e is satisficing. Essentially, satisficin g is a fo rm o f su b op tim izatio n . T h e re m ay b e a b e s t so lu tio n , a n o p tim u m , b u t it w o u ld b e difficult, i f n o t im p o ssib le, to attain it. W ith a n orm ative m o d el, to o m u ch co m ­ pu tatio n m ay b e involved ; w ith a d escrip tiv e m o d e l, it m ay n o t b e p o ss ib le to e v alu ate all th e sets o f alternatives.

R e la te d to s a tisficin g is S im o n ’s id e a o f b o u n d e d ration ality . H u m an s h a v e a lim ite d c a p a c ity fo r ra tio n a l th in k in g ; th e y g e n e r a lly c o n s tr u c t a n d a n a ly z e a sim ­ p lifie d m o d e l o f a re a l situ a tio n b y c o n s id e r in g fe w e r a lte rn a tiv e s, crite ria , and/or co n s tr a in ts th a n a c tu a lly e x is t. T h e ir b e h a v io r w ith r e s p e c t to th e s im p lifie d m o d e l m ay b e ra tio n a l. H o w e v e r, th e ra tio n a l s o lu tio n fo r th e s im p lifie d m o d e l m ay n o t b e ra tio n a l fo r th e re a l-w o rld p ro b le m . R a tio n a lity is b o u n d e d n o t o n ly b y lim ita tio n s o n h u m a n p r o c e s s in g c a p a c itie s , b u t a ls o b y in d iv id u a l d iffe r e n c e s , s u c h a s a g e , e d u c a ­ tio n , k n o w le d g e , an d attitu d e s. B o u n d e d ra tio n a lity is a ls o w h y m a n y m o d e ls are d e s crip tiv e ra th e r th a n n o rm a tiv e . T h is m a y a ls o e x p la in w h y s o m a n y g o o d m a n a g e rs re ly o n in tu itio n , a n im p o rta n t a s p e c t o f g o o d d e c is io n m a k in g ( s e e Stew art, 2 0 0 2 ; and P au ly, 2 0 0 4 ).

B e c a u s e ratio n ality a n d th e u s e o f n o rm a tiv e m o d e ls le a d to g o o d d e c is io n s , it is natu ral to a s k w h y s o m a n y b a d d e c is io n s a re m a d e in p ra c tic e . In tu itio n is a critical fa c to r th at d e c is io n m a k e rs u s e in so lv in g u n stru ctu re d a n d s e m istru ctu red p ro b le m s . T h e b e s t d e c is io n m a k e rs re c o g n iz e th e tr a d e -o ff b e tw e e n th e m arg in al c o s t o f o b ta in ­ ing fu rth e r in fo rm a tio n a n d a n aly sis v e rsu s th e b e n e fit o f m a k in g a b e tte r d e c is io n . B u t s o m e tim e s d e c is io n s m u st b e m a d e q u ick ly , an d , id e a lly , th e in tu itio n o f a s e a s o n e d , e x c e lle n t d e c is io n m a k e r is c a lle d fo r. W h e n a d e q u a te p la n n in g , fu n d in g , o r in fo rm a­ tio n is n o t a v a ila b le , o r w h e n a d e c is io n m a k e r is in e x p e r ie n c e d o r ill tra in e d , d isaste r c a n strik e.

Developing (Generating) Alternatives A significant part o f th e m o d el-bu ild in g p ro ce s s is g e n eratin g alternatives. In op tim ization m o d els (s u c h as lin ear program m ing), th e alternatives m ay b e g e n e ra te d au tom atically b y the m o d el. In m o st d ecisio n situations, h ow ev er, it is n e ce s s a ry to g e n era te alternatives m anually. T h is c a n b e a len gth y p ro ce s s that in v olv es search in g a n d creativity, p erh ap s utilizing e le ctro n ic b rainsto rm ing in a GSS. It ta k e s tim e and co s ts m o n ey . Issu e s su ch as w h e n to sto p g en eratin g alternatives c a n b e very im portant. T o o m an y alternatives c a n b e detrim ental to th e p ro cess o f d e cisio n m aking. A d e c is io n m a k e r m ay su ffer fro m inform a­ tio n overload .

G e n e ra tin g alternativ es is h e av ily d e p e n d e n t o n th e availability an d c o s t o f inform a­ tio n a n d re q u ires e x p e rtis e in th e p ro b le m are a. T h is is th e least form al a s p e c t o f problem solving. A lternatives c a n b e g e n e r a te d a n d e v a lu a te d u sin g h eu ristics. T h e g e n e ra tio n o f alternatives fro m e ith er individuals o r g ro u p s c a n b e s u p p o rte d b y e le c tro n ic b rain sto rm ­ ing softw are in a W e b -b a s e d GSS.

N ote th at th e s e a rch fo r a lternativ es u su ally o c c u rs after th e criteria fo r e valu atin g the alternativ es are d eterm in ed . T h is s e q u e n c e c a n e a se th e s e a r c h fo r a lternativ es and re d u ce th e e ffo rt in v o lv ed in evalu atin g th em , b u t identifying p o ten tial alternatives c a n so m e tim e s aid in id entify ing criteria.

Chapter 2 • Foundations and T ech n o lo g ies for D ecisio n Making

T h e o u tc o m e o f e v ery p ro p o s e d altern ativ e m u st b e estab lish ed . D e p e n d in g o n w h e th e r th e d e cisio n -m a k in g p ro b le m is classified a s o n e o l certainty, risk, o r u n certainty, d ifferent m o d e lin g a p p ro a ch e s m ay b e u se d (s e e D ru m m ond , 2 0 0 1 ; an d K o ller, 2 0 0 0 ). T h e se a re d iscu sse d in C h ap ter 9.

Measuring Outcomes T h e v a lu e o f a n altern ativ e is e v a lu a te d in te rm s o f g o a l attain m en t. S o m e tim e s an o u tc o m e is e x p r e s s e d d irectly in te rm s o f a g o a l. F o r e x a m p le , p ro fit is a n o u tc o m e , p ro fit m a x im iz a tio n is a g o a l, a n d b o th a re e x p r e s s e d in d o lla r term s. An o u tc o m e s u ch as cu sto m e r s a tisfa ctio n m a y b e m e a su re d b y th e n u m b e r o f co m p la in ts, b y th e lev el o f loyalty to a p ro d u ct, o r b y ratin gs fo u n d th ro u g h su rveys. Id ea lly , a d e c is io n m a k e r w o u ld w a n t to d e a l w ith a sin g le g o a l, b u t in p ra ctice , it is n o t u n u su a l to h a v e m u ltip le g o als (s e e B a rb a -R o m e ro , 2 0 0 1 ; an d K o k sa la n an d Z io n ts, 2 0 0 1 ). W h e n g ro u p s m ak e d e cisio n s, e a c h g ro u p p articip an t m ay h av e a d iffe re n t ag e n d a . F o r e x a m p le , e x e c u tiv e s m ight w a n t to m a x im iz e p ro fit, m ark etin g m ight w a n t to m a x im iz e m a rk e t p e n e tra tio n , o p era tio n s m ig h t w a n t to m in im ize c o s ts , an d s to c k h o ld e rs m ight w a n t to m a x im iz e the b o tto m lin e . T y p ica lly , th e s e g o a ls co n flict, s o s p e c ia l m u ltip le-criteria m e th o d o lo g ie s h av e b e e n d e v e lo p e d to h a n d le th is. O n e s u c h m e th o d is th e AHP. W e w ill stud y AHP in C h ap te r 9-

Risk All d e cis io n s are m a d e in a n in h eren tly u n stab le e n v iro n m en t. T h is is d u e to th e m any u n p re d icta b le e v e n ts in b o th th e e c o n o m ic a n d p h ysical en v iron m en ts. S o m e risk (m e a s ­ ured as p ro b ab ility ) m ay b e d u e to internal organ izatio n al e v e n ts, s u ch a s a valu ed e m p lo y e e q u ittin g o r b e c o m in g ill, w h e re a s o th ers m ay b e d u e to natu ral d isasters, s u ch as a h u rrican e. A side fro m th e h u m an toll, o n e e c o n o m ic a s p e c t o f H u rricane Katrina w as th at th e p rice o f a g allo n o f g a so lin e d o u b le d o v ern ig h t d u e to u n certain ty in th e port cap ab ilities, re fin in g, an d p ip e lin e s o f th e so u th e rn U n ited States. W h a t c a n a d e cisio n m ak e r d o in th e f a c e o f s u c h instability?

In g e n e r a l, p e o p le h a v e a te n d e n c y to m e a s u re u n c e rta in ty an d risk b a d ly . Purd y ( 2 0 0 5 ) said th a t p e o p le te n d to b e o v e r c o n fid e n t a n d h a v e a n illu s io n o f c o n tr o l in d e c is io n m a k in g . T h e re s u lts o f e x p e rim e n ts b y A d am G o o d ie a t th e U n iv ersity o f G e o rg ia in d ic a te th a t m o st p e o p le a re o v e r c o n fid e n t m o s t o f th e tim e (G o o d ie , 2 0 0 4 ). T h is m ay e x p la in w h y p e o p le o fte n f e e l th a t o n e m o re p u ll o f a s lo t m a c h in e w ill d e fin ite ly p a y o ff.

H ow ever, m e th o d o lo g ie s fo r h an d lin g e x tre m e u n certain ty d o exist. F o r e x a m p le , Y a k o v ( 2 0 0 1 ) d e s c rib e d a w a y to m a k e g o o d d ecisio n s b a s e d o n very little in fo rm atio n , using a n in fo rm a tio n g a p th e o ry an d m e th o d o lo g y ap p ro a ch . A side fro m estim atin g th e p o ten tial utility o r valu e o f a p articu lar d e c is io n ’s o u tc o m e , th e b e s t d e c is io n m a k e rs are c a p a b le o f accu ra te ly estim atin g th e risk a s s o cia te d w ith th e o u tc o m e s th at resu lt fro m m ak in g e a c h d e cis io n . T h u s, o n e im portant ta sk o f a d e c is io n m a k e r is to attrib u te a lev el o f risk to th e o u tc o m e a sso cia ted w ith e a c h p o ten tial alternative b e in g co n s id e re d . So m e d ecisio n s m ay le a d to u n a c c e p ta b le risks in term s o f s u c c e s s a n d c a n th e re fo r e b e dis­ card e d o r d is co u n te d im m ediately.

In s o m e c a s e s , s o m e d e c is io n s a re a ssu m e d to b e m a d e u n d e r c o n d itio n s o f c e r ­ tain ty sim p ly b e c a u s e th e e n v iro n m e n t is a ssu m e d to b e s ta b le . O th e r d e c is io n s are m ad e u n d e r c o n d itio n s o f u n certa in ty , w h e re risk is u n k n o w n . Still, a g o o d d e c is io n m a k e r c a n m a k e w o rk in g e stim a tes o f risk. A lso, th e p r o c e s s o f d e v e lo p in g BI/DSS in v o lv e s le a rn in g m o re a b o u t th e situ atio n , w h ic h le a d s to a m o re a c c u r a te a s s e s s m e n t o f t h e risks.

8 4 Part I • D ec isio n Making and Analytics: An O verview

Scenarios A scen ario is a s ta te m e n t o f a ssu m p tio n s a b o u t th e o p e ra tin g e n v iro n m e n t o f a p articu ­ lar s y stem a t a g iv e n tim e ; th a t is, it is a narrative d e s crip tio n o f th e d ecisio n -situ a tio n settin g. A s c e n a rio d e s c rib e s th e d e c is io n an d u n c o n tro lla b le v a ria b le s a n d p a ia m e te rs fo r a s p e c ific m o d e lin g situ ation . It m a y a lso p ro v id e th e p ro c e d u re s a n d co n strain ts for

th e m o d e lin g . . S ce n a rio s originated in th e theater, a n d th e te rm w a s b o rro w e d for w ar g am in g and

larg e-scale sim ulations. Sce n ario p lan n in g and an aly sis is a D SS to o l that can cap tu re a w h o le range o f possibilities. A m an ag er c a n co n stru ct a serie s o f sce n a rio s (i.e ., w hat-ir c a s e s ), p erfo rm co m p u terized analyses, a n d learn m o re a b o u t th e system a n d d ecisio n ­ m ak in g p ro b le m w h ile analyzing it. Ideally, th e m a n a g e r can identify an e x c e lle n t, p o ssibly op tim al, solu tion to th e m o d el o f th e p ro blem .

S ce n a rio s a re e sp e cia lly h e lp fu l in sim u latio ns a n d w h a t-if an aly ses. In b o th ca ses, w e c h a n g e sce n a rio s a n d e x a m in e th e results. F o r e x a m p le , w e c a n c h a n g e th e an ticip ated d em an d fo r h osp italizatio n (a n input v ariab le fo r p la n n in g ), thu s creatin g a n e w scen ario . T h e n w e c a n m e asu re th e a n ticip ated ca sh flo w o f th e h osp ital fo r e a c h scen ario .

S ce n a rio s p lay a n im portant ro le in d e c is io n m ak in g b e c a u s e they:

• H elp id en tify o p p o rtu n ities and p ro b le m areas • P rov id e flex ibility in p lan n in g • Id en tify th e lead in g e d g e s o f c h a n g e s that m an a g e m e n t sh o u ld m o n ito r • H e lp validate m ajo r m o d e lin g assu m p tio n s • A llow th e d e c is io n m a k e r to e x p lo r e th e b e h a v io r o f a sy stem th ro u g h a m o d el • H elp to c h e c k th e sensitivity o f p ro p o s e d so lu tio n s to c h a n g e s in th e en viron m en t,

as d e s crib e d b y th e scen a rio

Possible Scenarios T h e re m ay b e th o u san d s o f p o ss ib le s ce n a rio s fo r e v e iy d e c is io n situation. H o w ev er, th e

fo llo w in g are e sp e cia lly u sefu l in p ractice:

• T h e w o rst p o ss ib le scen a rio • T h e b e s t p o ss ib le scen a rio • T h e m o st lik e ly scen a rio • T h e a v e ra g e s ce n a rio

T h e sce n a rio d eterm in e s th e c o n te x t o f t h e analysis t o b e p erfo rm ed .

Errors in Decision Making T h e m o d e l is a critical c o m p o n e n t in th e d ecisio n -m a k in g p ro c e s s , b u t a d e c is io n m aker m ay m a k e a n u m b e r o f errors in its d ev e lo p m e n t a n d u se. V alidating th e m o d e l b e fo re it is u s e d is critical. G ath e rin g th e right am o u n t o f in fo rm atio n , w ith th e right lev el o f p re ci­ s io n a n d a ccu ra cy , to in co rp o ra te in to th e d e cisio n -m a k in g p ro c e s s is a lso critical. Saw yer ( 1 9 9 9 ) d e s crib e d “th e s e v e n d ead ly sin s o f d e c is io n m a k in g ,” m o st o f w h ic h are b eh av io r

o r in fo rm atio n related .

SECTION 2 . 5 REVIEW QUESTIONS

1 . D e fin e op tim ization and co n trast it w ith su boptim ization . 2 . C o m p are th e n orm ative a n d d escrip tive a p p r o a c h e s to d e c is io n m aking. 3 . D e fin e r a tio n a l d ec isio n m akin g . W h a t d o e s it really m e a n to b e a rational d ecisio n

m aker? 4 . W h y d o p e o p le e x h ib it b o u n d e d rationality w h e n solv in g problem s?

Chapter 2 • Foundations and T ech n o lo g ies for D ecisio n Making 8 5

6 S o m e “e rro rs” in d e c is io n m ak in g c a n b e attributed to th e n o tio n o f d e c is io n m aking fro m th e gut. E x p la in w h at is m e a n t b y this a n d h o w s u ch erro rs ca n h a p p e n .

5 . D e f in e sce n a rio . H o w is a s c e n a r io u s e d in d e c i s io n m a k in g ?

2.6 D ECISIO N M AKIN G : THE CHOICE PH ASE I n o ic e is th e critical a c t o f d e c is io n m aking. T h e c h o ic e p h a s e is th e o n e in w h ic h th e actual d e cisio n a n d th e co m m itm e n t to fo llo w a ce rta in co u rse o f a c tio n are m ad e . T h e -vDundary b e tw e e n th e d esig n a n d c h o ic e p h a s e s is o fte n u n cle a r b e c a u s e ce rta in activi­ tie s c a n b e p e rfo rm e d during b o th o f th e m an d b e c a u s e th e d e c is io n m a k e r c a n return freq u en tly fro m c h o ic e activities to d esig n activities (e .g ., g e n e ra te n e w altern ativ es w h ile perform ing a n e v a lu a tio n o f e x istin g o n e s ). T h e c h o ic e p h a s e in clu d es th e s e a r c h for, e v alu atio n o f, a n d re co m m e n d a tio n o f a n a p p ro p riate so lu tio n to a m o d el. A s o lu tio n to a m o d el is a s p e c ific s e t o f valu es fo r the d e c is io n v ariab les in a s e le c te d alternativ e. C h o ice s ca n b e e v alu ate d a s to th e ir viability a n d profitability.

N ote th at so lv in g a m o d e l is n o t th e sam e a s solv in g th e p ro b le m th e m o d e l re p re sen ts. T h e solu tion to th e m o d e l y ield s a re co m m e n d e d so lu tio n to th e p ro b lem . T h e p ro b le m is co n sid ered so lv e d o n ly i f th e re co m m e n d e d so lu tio n is s u cce ssfu lly im p lem en ted .

Solving a d ecisio n -m a k in g m o d e l involves s e a rch in g fo r a n a p p ro p ria te co u rse o f actio n . S e a rch a p p ro a ch e s in clu d e a n a ly tic a l te ch n iq u e s ( i.e ., s o l v i n g a form u la) a lg o rith m s ( i.e ., s te p -b y -ste p p ro ce d u re s ), h eu ristics (i.e ., ru les o f th u m b ), a n d b lin d sea rch es (i.e ., sh o o tin g in th e dark, ideally in a lo g ica l w a y ). T h e s e a p p r o a c h e s are

exam in ed in C h ap ter 9- . , E a c h a lte rn a tiv e m u st b e e v a lu a te d . I f a n altern ativ e h a s m u ltip le g o a ls , th e y m u st

all b e e x a m in e d a n d b a la n c e d a g a in st e a c h o th e r. Sensitivity analysis is u s e d to d e te r­ m in e th e r o b u s tn e s s o f a n y g iv e n altern ativ e; slig h t c h a n g e s in th e p a ra m e te rs sh o u ld ideally le a d to slig h t o r n o c h a n g e s in th e a ltern a tiv e ch o s e n . W hat-if an alysis is used to e x p lo r e m a jo r c h a n g e s in th e p a ra m e te rs . G o a l s e e k in g h e lp s a m a n a g e r d eter­ m in e v a lu e s o f th e d e c is io n v a r ia b le s to m e e t a s p e c ific o b je c tiv e . All this is d is cu ss e d

in C h ap ter 9.

SECTION 2 . 6 REVIEW QUESTIONS

1 . E x p la in th e d iffe re n ce b e tw e e n a p rin cip le o f c h o ic e an d th e actu al c h o ic e p h a se o f

d e cisio n m akin g. 2 . W h y d o s o m e p e o p le cla im that th e c h o ic e p h a s e is th e p o in t in tim e w h e n a d e cisio n

is really m ade? 3. H o w c a n sensitivity analysis h e lp in th e c h o ic e phase?

2.7 D ECISIO N M AKIN G : THE IM PLEM ENTATION PHASE In The P rince, M achiavelli astutely n o te d so m e 5 0 0 y ears a g o that th ere w a s ‘ n o th in g m o re difficult to carry ou t, n o r m o re d oubtful o f su cce ss, n o r m o re dangerou s to h an d le than to initiate a n e w ord e r o f things.” T h e im p lem entation o f a p ro p o se d solu tion to a p ro b le m is, in effe ct th e initiation o f a n e w ord er o f things o r th e introd u ction o f ch an g e. A nd ch an ge m ust b e 'm a n a g e d . U ser ex p e cta tio n s m ust b e m an ag ed as part o f ch an g e m anagem ent.

T h e d e fin itio n o f im p lem en tation is s o m e w h a t c o m p lic a te d b e c a u s e im p le m e n ta tio n is a lo n g in v o lv e d p ro ce s s w ith v a g u e b o u n d a rie s. Sim p listically, th e im plem entation p h a s e in v olv es p u ttin g a re c o m m e n d e d s o lu tio n to w o rk , n o t n e c e s s a rily im p lem e n tin g a co m p u te r sy stem . M any g e n e r ic im p le m e n ta tio n issu es, s u ch a s re s is ta n c e to change^ d e g r e e o f s u p p o rt o f to p m a n a g e m e n t, an d u s e r training , are im p ortan t in d e a lin g w ith

8 6 Part I • D ec isio n Making and Analytics: An O verview

in fo rm a tio n sy stem s u p p o rte d d e c is io n m ak in g . In d e e d , m a n y p re v io u s te c h n o lo g y - re la ted w a v e s (e .g ., b u sin e ss p r o c e s s re e n g in e e rin g (B P R ), k n o w le d g e m an ag e m e n t, e tc .) h a v e fa c e d m ix e d resu lts m ain ly b e c a u s e o f c h a n g e m a n a g e m e n t ch a lle n g e s and issu es. M a n a g e m e n t o f c h a n g e is alm o st a n e n tire d is c ip lin e in itself, s o w e r e c o g n iz e its im p o rta n ce an d e n c o u ra g e th e re a d ers to fo c u s o n it in d e p e n d en tly . Im p le m en ta tio n also in clu d e s a th o ro u g h u n d e rstan d in g o f p r o je c t m a n a g e m e n t. Im p o rta n ce o f p r o je c t m a n ­ a g e m e n t g o e s far b e y o n d a n a ly tics, so th e la st fe w y e a rs h av e w itn e sse d a m a jo r g ro w th in ce rtifica tio n p ro gram s fo r p r o je c t m an ag ers. A v e ry p o p u la r ce rtifica tio n n o w is P ro ject M an ag e m e n t P ro fe s s io n a l (P M P ). S e e p m i . o r g fo r m o re details.

Im p le m en ta tio n m u st a lso involve c o lle c tin g an d analyzing d ata to learn fro m th e prev iou s d ecisio n s an d im p ro v e th e n e x t d ecisio n . A lth o u g h analysis o f d ata is usu ally co n d u cte d to id entify th e p ro b le m and/or t h e solu tio n , an aly tics sh o u ld a lso b e e m p lo y e d in th e fe e d b a c k p ro ce ss. T h is is e sp e cia lly t a i e fo r an y p u b lic p o lic y d e cisio n s. W e n e e d to b e su re that th e d ata b e in g u s e d fo r p ro b le m id e n tificatio n is valid. S o m etim e s p e o p le find this o u t o n ly a fte r th e im p lem e n tatio n p h ase.

T h e d ecisio n -m a k in g p ro c e s s , th o u g h c o n d u cte d b y p e o p le , c a n b e im p ro v ed w ith co m p u te r su p p ort, w h ich is th e s u b je c t o f th e n e x t sectio n .

SECTION 2 . 7 REVIEW QUESTIONS

1 . D e fin e im plem entation. 2 . H o w ca n D SS su p p o rt th e im p lem e n tatio n o f a d ecisio n ?

2.8 H O W D E C IS IO N S A R E SU PPO R T ED In C h a p te r 1, w e d is cu ss e d th e n e e d fo r co m p u te riz e d d e c is io n s u p p o rt an d b riefly d e s c rib e d s o m e d e c is io n aids. H e re w e re la te s p e c ific te c h n o lo g ie s to th e d e c is io n ­ m a k in g p r o c e s s ( s e e F ig u re 2 .2 ). D a ta b a se s , d ata m arts, a n d e s p e c ia lly d ata w a r e h o u s e s are im p o rta n t te c h n o lo g ie s in su p p o rtin g all p h a s e s o f d e c is io n m a k in g . T h e y p ro v id e th e d ata that drive d e c is io n m ak in g .

Support for the Intelligence Phase T h e prim ary requ irem en t o f d ecisio n supp ort for th e in tellig en ce p h ase is th e ability to scan e xte rn al and internal inform ation so u rces fo r op p ortu nities an d p ro blem s a n d to interpret w h at th e scan n in g discovers. W e b to o ls an d so u rce s are e x trem ely usefu l fo r environm ental

P h ase ANN MIS

Data Mining, OLAP E S, E R P

E S S , E S , SCM CRM, ERP, K V S Management Science ANN

D S S E S

E S S , E S KMS, E R P

CRM SCM

FIG U R E 2 .2 DSS Support.

Chapter 2 • Foundations and T ech n o lo g ies fo r D ec isio n Making

scanning. W e b b ro w se rs p rovide u sefu l front en d s fo r a variety o f to ols, fro m OLAP to data m ining and data w areh o u ses. D ata so u rces c a n b e internal o r external. Internal so u rce s may b e a cce ssib le via a co rp o ra te intranet. E xternal so u rces are m any an d varied.

D e c is io n su p p ort/ B I te c h n o lo g ie s c a n b e v e ry h elp fu l. F o r e x a m p le , a d ata w a r e ­ h o u s e c a n su p p o rt th e in te llig e n c e p h a s e b y co n tin u o u sly m o n ito rin g b o th in te rn al a n d e x te rn a l in fo rm atio n , lo o k in g fo r e a rly sig n s o f p ro b le m s and o p p o rtu n itie s th ro u g h a W e b -b a s e d e n te rp ris e in fo rm a tio n p o rtal (a ls o ca lle d a d a s h b o a rd ). Sim ilarly, (a u to m a tic) d ata (a n d W e b ) m in in g (w h ic h m ay in clu d e e x p e rt sy stem s [ES], CRM, g e n e tic alg o rith m s, n eu ral n e tw o rk s, a n d o th e r an aly tics s y ste m s) a n d (m a n u a l) OLAP a ls o su p p o rt th e in te l­ lig e n ce p h a s e b y id e n tify in g re la tio n sh ip s am o n g activ ities an d o th e r facto rs. G e o g ra p h ic inform ation sy stem s (G IS ) ca n b e u tilized e ith e r a s s ta n d -a lo n e sy stem s o r in te g ra te d w ith th e se sy stem s s o th a t a d e c is io n m a k e r c a n d ete rm in e o p p o rtu n itie s and p ro b le m s in a spatial s e n s e . T h e s e re la tio n sh ip s c a n b e e x p lo ite d fo r co m p etitiv e ad v an tag e ( e .g ., CRM identifies cla ss e s o f cu sto m e rs to a p p r o a c h w ith s p e c ific p ro d u cts an d s e r v ic e s ). A KMS c a n b e u s e d to id en tify sim ilar p a st situ atio n s an d h o w th e y w e re h an d led . G S S c a n b e u se d to s h a re in fo rm a tio n and fo r b rain sto rm in g . As s e e n in C h ap te r 14, e v e n c e ll p h o n e and G P S d ata c a n b e ca p tu re d to c re a te a m icro -v ie w o f cu sto m e rs an d th e ir h ab its.

A n oth er a s p e c t o f id entifying in tern al p ro b lem s and cap ab ilitie s in v olv es m o n ito rin g d ie cu rren t statu s o f o p era tio n s. W h e n so m e th in g g o e s w ro n g , it c a n b e id en tified q uickly and th e p ro b le m c a n b e so lv ed . T o o ls s u ch as b u sin e ss activity m o n ito rin g (B A M ), b u si­ n ess p ro ce s s m an a g e m e n t (B P M ), an d p ro d u ct life -cy cle m an a g e m e n t (PLM) p ro v id e su ch cap ab ility to d e c is io n m ak ers. B o th ro u tin e and a d h o c rep o rts c a n aid in th e in te llig e n ce p h ase. F o r e x a m p le , regu lar rep o rts c a n b e d esig n ed to assist in th e p ro b lem -fin d in g activity b y co m p a rin g e x p e c ta tio n s w ith cu rre n t an d p ro je c te d p e rfo rm an ce . W e b -b a s e d OLAP to o ls are e x c e lle n t at this task. So a re visu alization to o ls a n d e le c tro n ic d o cu m en t m an ag em en t system s.

E xp e rt sy stem s (E S ), in con trast, c a n re n d er a d v ice reg ard ing th e n atu re o f a p ro b ­ lem , its cla ssifica tio n , its serio u sn e ss, an d th e like. ES c a n ad vise o n th e su itab ility o f a solution a p p r o a c h a n d th e lik e lih o o d o f s u cce ssfu lly solv in g th e p ro b lem . O n e o f th e primary are a s o f ES s u c c e s s is in terp retin g in form ation a n d d iag n o sin g p ro b le m s. T h is capability c a n b e e x p lo ite d in th e in te llig e n ce p h a s e . E v e n in tellig en t ag en ts c a n b e u sed 10 identify op p o rtu n ities.

M u ch o f th e in form ation u s e d in s e e k in g new' o p p o rtu n ities is q u alitativ e, o r soft. This in d icates a h ig h lev el o f u n stru ctu red n ess in th e p ro b le m s, thu s m ak in g D SS q u ite useful in th e in te llig e n ce p h ase.

T h e In te rn e t a n d a d v a n ce d d atab ase te c h n o lo g ie s h a v e c re a te d a glut o f d ata and inform ation a v ailab le to d e c is io n m ak ers— s o m u ch that it c a n d etract fro m th e quality and s p e e d o f d e c is io n m aking. It is im portant to re co g n iz e s o m e issu es in u sin g d ata and analytics to o ls fo r d e c is io n m akin g. First, to p arap h rase b a s e b a ll g reat V in Scully, “data should b e u se d th e w a y a d run k u s e s a lam p p o st. F o r su p p ort, n o t fo r illu m in atio n .” It ^ e sp ecia lly true w h e n th e fo cu s is o n u n d erstan d in g th e p ro b lem . W e sh o u ld re co g n iz e tftat n o t all th e data th a t m ay h elp un derstand th e p ro b le m is a v ailab le. T o q u o te Einstein, ■“Not everyth in g th a t co u n ts c a n b e co u n te d , a n d n o t everyth in g that c a n b e co u n te d cou nts.” T h e re m ig h t b e o th e r issu es th at h av e to b e re co g n iz e d as w ell.

Support fo r the Design Phase H i e d esig n p h a s e in v olv es g e n era tin g alternative co u rs e s o f a ctio n , d iscu ssin g th e criteria far c h o ic e s a n d th e ir relativ e im p o rtan ce , and fo re ca stin g th e future c o n s e q u e n c e s o f ■«ang v arious altern ativ es. S ev eral o f th e s e activities ca n u s e stand ard m o d els p ro v id e d b y a D SS (e .g ., fin an cial an d fo reca stin g m o d e ls, a v ailab le as a p p le ts). A lternatives fo r stru c- c_red p ro b lem s c a n b e g e n e ra te d th ro u g h th e u s e o f e ith e r stand ard o r s p e c ia l m o d els.

H o w e v e r th e g e n era tio n o f a lternativ es fo r c o m p le x p ro b le m s req u ires e x p e rtis e that c a n b e p ro v id e d o n ly b y a hu m an, b rain sto rm in g softw are, o r a n ES. OLAP an d d ata m m m g softw are are q u ite u sefu l in identifying re latio n sh ip s th a t c a n b e u se d m m od els. M ost DSS h av e quantitative analysis cap ab ilitie s, and a n internal ES c a n assist w ‘th qUal‘ta” e " od s as w e ll as w ith th e e x p e rtis e re q u ired in s e le ctin g q u an titative analysis an d fo re ca stin g m o d e ls A KMS sh o u ld ce rtain ly b e co n su lte d to d eterm in e w h e th e r s u ch a p ro b le m has b e e n e n c o u n te re d b e fo r e o r w h e th e r th e re are e x p e rts o n h an d w h o ca n u n d erstan d in g a n d an sw ers. CRM sy stem s, re v e n u e m a n a g e m e n t sy stem s, ERP, and K , system s softw are a r e u sefu l in that th e y p ro v id e m o d e ls o f b u s i n e s s p ro c e s s e s th a t can ^ e s assu m p tio n s an d scen a rio s. I f a p ro b le m re q u ires b rain sto rm in g to h e lp id en tify im portant issu es an d o p tio n s, a G SS m ay p ro v e h elp fu l. T o o ls th a t pro v id e co g n itiv e m apping; can a lso h e lp C o h e n e t al. (2 0 0 1 ) d e s crib e d sev eral W e b -b a s e d to o ls that p rovide d e cisio n su p p ort, m ainly in the d esig n p h a s e , b y providing m o d e ls an d re p o rtin g o f alternative results E a c h o f th e ir ca s e s has sa v e d m illions o f d ollars an n u ally b y utilizing th e s e to ols. Su ch D SS are h e lp in g e n g in e e rs in p ro d u ct d esig n as w e ll as d e c is io n m ak ers solving

b u sin e ss p ro b lem s.

Support for the Choice Phase In ad dition to providing m o d els that rapidly identify a b e s t o r g o o d -en o u g h alternative a D SS c a n supp ort th e c h o ic e p h ase throu gh w h at-if an d go al-seek m g analyses. D ifferent scen ario s can b e tested fo r th e sele cte d o p tio n to re in fo rce th e final decision. Again, a KMS h elp s identify sim ilar past e x p e rien ces; CRM, ERP, an d SCM system s are u se d to test th e im pacts o f d ecisio n s in establishing their value, lead ing to an intelligent ch o ice . An ES can b e u sed to assess th e desirability o f ce rta in solutions as w e ll as to re co m m en d an ap p ro p n - ate solution. If a group m ak e s a d ecisio n , a G SS c a n p rovide supp ort to lead to consensus.

Support fo r the Im plem entation Phase T h is is w h e re “m ak in g th e d e c is io n h a p p e n ” o ccu rs. T h e D SS b e n e fits pro v id ed during im p lem e n ta tio n m ay b e as im p ortan t as o r e v e n m o re im p o rtan t th a n th o se in th e e a r i p h a se s. D SS c a n b e u s e d in im p lem e n tatio n activities s u ch a s d e c is io n co m m u n icatio n ,

ex p la n a tio n , a n d ju stification . f Im p le m en ta tio n -p h a se D SS b e n e fits a re partly d u e to the vivid ness an d d etail o f

an aly ses and rep orts. F o r e x a m p le , o n e c h ie f e x e c u tiv e o ffice r (C E O ) gives e m p lo y e e s and e xte rn al p arties n o t o n ly th e a g g re g ate fin an cial g o a ls and c a s h n e e d s fo r th e n e a r term b u t a lso th e calcu latio n s, interm ed iate results, an d statistics u sed m d eterm in g th e ag g re g ate figures. In ad d ition to co m m u n icatin g th e fin an cial g o als u n am b ig u o u sly th e C E O sign als o th er m e ssa g es. E m p lo y e e s k n o w th a t th e CEO has th o u g ro“ 8 . assu m p tio n s b e h in d th e fin an cial g o als an d is serio u s a b o u t th eir im p o rtan ce an d attain­ ability B a n k e rs a n d d irecto rs are sh o w n th at th e C E O w as p e rso n ally i n v o k e d m a n a ­ lyzing c a s h n e e d s and is a w a re o f an d re s p o n s ib le fo r th e im p lications o f th e financing re q u e sts p re p a re d b y th e fin an ce d ep artm en t. E a c h o f th e se m e ssa g es im p ro ves d e cisio n

im p lem e n ta tio n in s o m e w ay . . , , As m e n tio n e d e arlie r, re p o rtin g sy stem s an d o th e r to o ls variou sly la b e le d a s B

BPM KMS EIS, ERP, CRM, an d SCM a re all u sefu l in tra ck in g h o w w ell an im p lem en tatio is w o rk in g . G SS is u sefu l fo r a te a m to c o lla b o ra te in estab lish in g im p lem e n tatio n e ffe c ­ tiv en ess. F o r e x a m p le , a d e c is io n m ight b e m a d e to g e t rid o f u n p ro fitab le cu sto m e is . An e ffe ctiv e CRM c a n id entify cla ss e s o f cu sto m ers to g e t rid o f, id entify th e im p act o f d oing

so a n d th e n v erify that it really w o rk e d th a t w ay. All p h ases o f the decision-m aking p ro cess c a n b e su p p orted b y im proved com m u nica­

tion through collaborative com puting via G SS and KMS. Com puterized system s ca n facilitate com m unication b y help ing p e o p le e xp lain and justify th e ir suggestions and opinions.

D ec isio n Making and Analytics: An Overview

Chapter 2 • Foundations and T ech n o lo g ies fo r D ecision Making

D e cis io n im p lem e n ta tio n c a n also b e s u p p o rte d b y ES. An ES c a n b e u s e d a s a n advi­ sory sy stem reg ard in g im p lem e n tatio n p ro b lem s (s u c h as h an d lin g re sista n ce to c h a n g e ). Finally, a n ES c a n p ro v id e training that m ay s m o o th th e c o u rs e o f im p lem en tatio n .

Im pacts alo n g th e v alu e ch ain , th o u g h rep o rted b y an EIS throu gh a W e b -b a s e d enterprise in form ation portal, are typically id entified b y BAM, BPM , SCM, a n d ERP system s. CRM system s rep ort a n d up d ate internal reco rd s, b ase d o n th e im p acts o f th e im p lem en ta- oon. T h e s e inputs a re th e n u s e d to identify n e w p ro b lem s an d o p p ortu nities— a return to d ie in tellig en ce p h ase.

SECTION 2 . 8 REVIEW QUESTIONS

1. D escrib e h ow D SS/BI tech n olog ies an d tools c a n aid in e a ch p h ase o f d ecisio n making.

2. D e s c r ib e h o w n e w te c h n o lo g ie s c a n pro v id e d ecisio n -m a k in g support.

Now th at w e h av e stu d ied h o w te c h n o lo g y c a n assist in d e cisio n m akin g, w e study so m e details o f d e c is io n su p p o rt sy stem s (D S S ) in th e n e x t tw o s ectio n s.

2,9 D EC ISIO N SU PPO R T S Y S T E M S : C A P A B IL IT IE S The early d efin itio n s o f a D SS id en tified it as a sy stem in te n d e d to su p p o rt m an ag erial d ecisio n m ak ers in sem istru ctu red an d unstructured d e c is io n situations. D SS w e re m e an t

: b e ad ju n cts to d e c is io n m ak e rs, e x te n d in g th eir c ap ab ilitie s b u t n o t re p la cin g th e ir ju d g­ m ent. T h e y w e r e a im e d a t d ecisio n s that re q u ired ju d g m e n t o r a t d e cis io n s th at co u ld n o t r e co m p le te ly su p p o rte d b y algorithm s. N ot sp e cifica lly stated b u t im plied in th e early d efinitions w as th e n o tio n that th e system w o u ld b e co m p u te r b a se d , w o u ld o p e ra te inter-

o n lin e, a n d p re fe rab ly w o u ld h a v e grap h ical ou tp u t ca p a b ilitie s, n o w sim p lified ~rz. b row sers an d m o b ile d ev ices.

- DSS Application D SS is typically b u ilt to su p p o rt th e so lu tio n o f a ce rta in p ro b le m o r to e v alu ate an

icp o rtu n ity . T h is is a k e y d iffe ren ce b e tw e e n D SS and B I ap p licatio n s. In a v e ry strict ease, business intelligence (B I ) system s m o n ito r situ ation s and id entify p ro b le m s and/

o p p ortu n ities, u sin g analytic m eth o d s. R e p o rtin g plays a m a jo r ro le in B I; th e u s e r g en erally m ust id entify w h e th e r a p articu lar situ ation w arrants atten tion , a n d th e n analyti-

i m eth o d s c a n b e ap p lied . A gain, alth o u g h m o d e ls an d data a c c e s s (g e n e ra lly throu gh d iia w a re h o u se ) a re in clu d e d in B I, D SS typically h av e th e ir o w n d a ta b a se s a n d are

E '-d o p e d to s o lv e a s p e c ific p ro b le m o r s e t o f p ro b lem s. T h e y a re th e re fo r e calle d 3SS applications.

Fo rm ally , a D S S is a n a p p r o a c h (o r m e th o d o lo g y ) fo r s u p p o rtin g d e c is io n m ak in g . k ^ 5 e s a n in te ra c tiv e , fle x ib le , a d a p ta b le c o m p u te r -b a s e d in fo rm a tio n s y s te m (C B IS ) s p e c ia lly d e v e lo p e d fo r s u p p o rtin g th e s o lu tio n to a s p e c ific u n stru c tu re d m a n a g e - m=-- p ro b le m . It u s e s d ata, p ro v id e s a n e a s y u s e r in te rfa c e , a n d c a n in c o r p o r a te th e a o sio n m a k e r’s o w n in sig h ts. In ad d itio n , a D SS in c lu d e s m o d e ls a n d is d e v e lo p e d

possibly b y e n d u s e r s ) th r o u g h a n in te ra c tiv e a n d itera tiv e p r o c e s s . It c a n s u p p o r t all o f d e c is io n m a k in g a n d m ay in c lu d e a k n o w le d g e c o m p o n e n t. F in a lly , a D SS

i b e u s e d b y a s in g le u s e r o r c a n b e W e b b a s e d fo r u s e b y m a n y p e o p le a t sev eral ( h c E l O O S .

B e c a u s e th e re is n o c o n s e n s u s o n e x a ctly w h at a D SS is, th ere is o b v io u sly n o a g re e- y - r o n th e stan d ard ch aracteristics an d cap ab ilitie s o f D SS. T h e cap ab ilitie s in F ig u re 2.3

in s t i t u t e a n id eal set, s o m e m e m b ers o f w h ic h are d e s crib e d in th e d efin itio n s o f D SS j o d illustrated in th e a p p licatio n ca ses.

T h e k e y ch a ra cteristics a n d cap ab ilitie s o f D SS (a s sh o w n in Figu re 2 .3 ) are:

9 0 Part I • D e c is io n Making and Analytics: An Overview

Support individuals

and groups

Modeling and analysis Interdependent

or sequential decisionsD ecision S u p p o rt

S y s t e m s (D SS}Ease of development by end users

Humans control the process Support variety

of decision processes and styles

Effectiveness and efficiency Interactive,

ease of use

Data access

Support intelligence

design, choice, and implementation

Adaptable and flexible

Stand-alone, integration, and

Web-based

Semistructured or unstructured

problems Support

managers at all levels

FIGURE 2.3 Key Characteristics and Capabilities of DSS.

1 . Su p p ort fo r d e c is io n m ak e rs, m ainly in sem istru ctu red a n d u n stru ctu red situations, b y b rin g in g to g e th e r h u m an ju d g m en t and co m p u te rize d in form ation. S u c h p ro b ­ lem s ca n n o t b e so lv e d (o r c a n n o t b e so lv e d c o n v e n ie n tly ) b y o th e r co m p u terized system s o r th ro u g h u s e o f stan d ard q u antitative m e th o d s o r to ols. G e n e rally , th e se p ro b lem s g ain structure as th e D SS is d e v e lo p e d . E v e n s o m e stru ctu red p ro b lem s

have b e e n so lv e d b y DSS. 2 . Su p p ort fo r all m an ag erial lev els, ran g in g fro m to p e x e c u tiv e s to lin e m anagers. 3 Su p p ort fo r individuals a s w e ll a s g ro u p s. L ess-stru ctu red p ro b lem s o fte n req u ire th e

in v olv em en t o f individuals fro m d iffe ren t d ep artm e n ts a n d org an izatio n al lev els or e v e n fro m d ifferent organizations. D SS su p p o rt virtual te am s th ro u g h co llab o rativ e W e b to o ls. D SS h av e b e e n d e v e lo p e d to su p p o rt individual a n d g ro u p w o rk , a s w ell a s to su p p o rt individ ual d e c is io n m ak in g an d g ro u p s o f d e c is io n m ak ers w o rk in g

so m e w h a t in d ep en d en tly. 4 . Su p p ort fo r in te rd e p e n d e n t and/or se q u e n tia l d e cisio n s. T h e d e cis io n s m ay b e m ade

o n c e , sev eral tim es, o r rep eated ly. 5 . Su p p ort in all p h a s e s o f th e d ecisio n -m a k in g p ro ce ss: in te llig e n ce , d esig n , ch o ic e ,

and im p lem en tation . 6 . S u p p o rt fo r a variety o f d ecisio n -m a k in g p ro c e s s e s an d styles. 7 . T h e d e c is io n m a k e r shou ld b e reactiv e, a b le to c o n fro n t ch a n g in g co n d itio n s q u ic k y,

and a b le to ad ap t th e D S S to m e e t th e s e ch a n g e s. D SS are fle x ib le , s o u se rs c a n add, d e le te , c o m b in e , ch a n g e , o r rearran g e b a s ic e le m e n ts . T h e y are a ls o fle x ib le in that th e y c a n b e read ily m o d ified to s o lv e o th er, sim ilar p ro b lem s.

C hapter 2 • Foundations a n d T ech n o lo g ies for D ecisio n Making

8 . U ser-frien d lin ess, stro n g g rap h ical cap ab ilitie s, an d a natu ral la n g u a g e interactive h u m a n -m a c h in e in te rface c a n greatly in c re a se th e e ffe ctiv e n e s s o f D SS. M o st n e w D SS ap p lica tio n s u s e W e b -b a s e d in te rfa ce s o r m o b ile p latfo rm in te rface s.

9 . Im p ro v e m e n t o f th e e ffe ctiv e n e s s o f d e c is io n m ak in g (e .g ., a ccu ra cy , tim elin ess, q u ality) ra th er th a n its e ffic ie n c y (e .g ., th e c o s t o f m ak in g d e c is io n s ). W h e n D SS are d e p lo y e d , d e cisio n m aking o fte n ta k e s lo n g e r, b u t th e d ecisio n s a re b etter.

10 . T h e d e c is io n m a k e r h a s co m p le te co n tro l o v e r all step s o f th e d ecisio n -m a k in g p ro c e s s in solv in g a p ro b lem . A D SS s p e cifica lly aim s to su p p ort, n o t to re p la ce , th e d e c is io n m ak er.

1 1 . E n d u se rs are a b le to d ev elo p and m o d ify s im p le system s b y th e m se lv es. Larger system s c a n b e b u ilt w ith a ssistan ce fro m in form ation sy stem (IS ) sp ecialists. S p re a d s h e e t p a c k a g e s h av e b e e n utilized in d ev elo p in g sim p le r sy stem s. OLAP and d ata m in in g so ftw are, in c o n ju n c tio n w ith data w a reh o u se s, e n a b le users to build fairly la rg e , c o m p le x DSS.

12 . M odels a re g e n era lly utilized to an aly ze d ecisio n -m a k in g situations. T h e m o d ­ e lin g cap a b ility e n a b le s e x p e rim e n ta tio n w ith d ifferen t strateg ies u n d e r d ifferen t c o n fig u ra tio n s.

1 3 . A cce ss is p ro v id e d to a variety o f d ata so u rce s, fo rm ats, an d ty p es, in clu d in g GIS, m u ltim ed ia, a n d o b je c t-o rie n te d data.

14 . T h e D SS can b e e m p lo y e d as a stand -alone to o l u se d b y an individual d ecisio n m ak er in o n e lo ca tio n o r distributed throughout a n organization and in sev eral organizations a lo n g th e supp ly ch ain . It c a n b e integrated w ith o th er D SS and/or ap p licatio n s, and it c a n b e distributed internally an d externally, u sin g netw orkin g an d W e b tech n olog ies.

T h e s e k e y D SS ch aracteristics and cap ab ilitie s allo w d e c is io n m a k e rs to m ake b etter, m o re c o n s is te n t d e cis io n s in a tim ely m ann er, and th e y are p ro v id e d b y the m ajor DSS co m p o n e n ts , w h ic h w e w ill d e scrib e a fte r d iscu ssin g vario u s w a y s o f classifying D SS (n e x t).

SECTION 2 . 9 REVIEW QUESTIONS

1 . List th e k e y ch aracteristics a n d cap a b ilitie s o f D SS. 2 . D e s c r ib e h o w p ro vid ing su p p ort to a w o rk g ro u p is d ifferen t fro m p ro v id in g su p p ort

to g ro u p w o rk . E x p la in w h y it is im portant to d ifferen tiate th e s e c o n c e p ts.

3 . W h at k in d s o f D SS c a n en d u se rs d ev elo p in sp read sh eets? 4. W h y is it s o im p ortan t to in clu d e a m o d e l in a DSS?

2.10 D S S C L A SS IF IC A T IO N S D SS ap p licatio n s h av e b e e n classified in sev eral d ifferent w ays (s e e P o w e r, 2 0 0 2 ; P o w er and Sharda, 2 0 0 9 ). T h e d esign p ro cess, as w e ll as th e o p eratio n and im p lem entation o f DSS, d ep en d s in m any cases o n th e typ e o f D SS involved. H ow ever, re m e m b er that n o t every DSS fits n eatly into o n e category. M ost fit into th e classification provid ed b y th e A ssociation fo r Info rm ation System s Sp ecial In terest G ro u p o n D e cis io n Su p p ort System s (AIS SIG D SS). W e discuss this classification b u t a lso point o u t a fe w o th er attem pts at classifying DSS.

The AIS SIGD SS Classification for DSS T h e AIS SIG D SS (ais.site-ym .com /group/SIG D SS) h a s a d o p te d a c o n c is e classificatio n s c h e m e fo r D SS th at w a s p ro p o s e d b y P o w e r (2 0 0 2 ). It in clu d e s th e fo llo w in g cate g o rie s:

• C o m m u n icatio n s-d riv en a n d g ro u p D SS (G SS) • D ata-d riv e n D SS

9 2 Part I • D ecisio n M aking and Analytics: An O verview

• D o cu m e n t-d riv e n D SS • K n o w led g e -d riv en D SS, data m ining , an d m a n a g e m e n t ES ap p licatio n s • M o d el-d riv en D SS

T h e r e m ay a ls o b e hybrid s th at co m b in e tw o o r m o re ca te g o rie s. T h e s e are calle d c o m p o u n d DSS. W e d iscu ss th e m a jo r ca te g o rie s n ext.

C O M M U N IC A T IO N S -D R IV E N A N D G R O U P D S S C o m m u n ica tio n s -d riv e n a n d g ro u p D SS (G S S ) in c lu d e D SS th a t u se c o m p u te r, co lla b o r a tio n , a n d c o m m u n ic a tio n te c h n o lo g ie s to s u p p o rt g ro u p s in ta s k s th a t m ay o r m ay n o t in c lu d e d e c is io n m a k in g . E ssen tially, all D SS th a t s u p p o rt an y k in d o f g ro u p w o rk fa ll in to this ca te g o ry . T h e y in clu d e th o se th at s u p p o rt m e e tin g s , d e s ig n co lla b o r a tio n , a n d e v e n su p p ly ch a in m a n a g e m e n t. K n o w le d g e m a n a g e m e n t sy stem s (K M S) that a re d e v e lo p e d a ro u n d co m m u n itie s that p ra c tic e c o lla b o r a tiv e w o rk a ls o fall in to th is ca te g o ry . W e d iscu ss th e s e in m o re d etail in la te r ch a p te rs.

D A T A -D R IV E N D S S D ata-d riv en D SS a re prim arily in v o lv e d w ith d ata a n d p ro cessin g th e m in to in form ation a n d p re sen tin g th e in fo rm atio n to a d e c is io n m ak er. M any DSS d e v e lo p e d in OLAP an d rep ortin g an alytics softw are sy stem s fall in to this cate g o ry . T h e re is m inim al em p h a sis o n th e u se o f m ath em atical m o d e ls.

In this ty p e o f D SS, th e d a ta b a s e o r g a n iz a tio n , o f te n in a d ata w a r e h o u s e , p lays a m a jo r r o le in th e D SS stru ctu re. E arly g e n e r a tio n s o f d a ta b a s e -o r ie n te d D SS m ain ly u s e d th e r e la tio n a l d a ta b a s e c o n fig u ra tio n . T h e in fo r m a tio n h a n d le d b y re la tio n a l d a ta b a s e s te n d s to b e v o lu m in o u s, d e s crip tiv e , a n d rigid ly stru ctu red . A d a ta b a s e - o r ie n te d D SS fe a tu re s s tro n g re p o rt g e n e r a tio n a n d q u e ry c a p a b ilitie s . In d e e d , this is p rim arily th e c u rre n t a p p lic a tio n o f th e to o ls m a rk e d u n d e r th e B I u m b re lla o r u n d e r th e la b e l o f re p o rtin g / b u sin e ss an a ly tics. T h e c h a p te r s o n d ata w a r e h o u s in g an d b u s in e s s p e rfo rm a n c e m a n a g e m e n t (B P M ) d e s c r ib e s e v e r a l e x a m p le s o f th is c a te g o ry o f D SS.

D O C U M EN T-D R IV EN D S S D o cu m en t-d riv en D SS re ly o n k n o w le d g e co d in g , analysis, search , and retrieval fo r d ecisio n supp ort. T h e y essen tially in clu d e all D SS that are text b a se d . M ost KMS fall in to this categ ory . T h e s e D SS a ls o h av e m inim al em p h asis o n utiliz­ ing m ath em atical m o d els. F o r e x a m p le , a system th at w e built fo r th e U.S. Army s D efen se A m m unitions C en ter falls in this category . T h e m ain o b je c tiv e o f d ocu m en t-d riven D SS is to provid e su p p ort fo r d e cisio n m ak in g u sin g d o cu m en ts in v arious form s: oral, w ritten, an d m ultim edia.

K N O W LE D G E -D R IV E N D S S , D A T A M IN IN G , A N D M A N A G E M E N T E X P E R T S Y S T E M S A P P L IC A T IO N S T h e s e D SS involve th e a p p licatio n o f k n o w le d g e te ch n o lo g ie s to address sp e cific d e c is io n supp ort n e ed s. Essentially, all artificial in te llig e n c e -b a s e d D SS fall into this categ ory . W h e n s y m b o lic storag e is utilized in a D SS, it is g en erally in this category. ANN a n d ES a re in clu d e d h e re . B e c a u s e th e b en efits o f th e s e intelligent DSS o r know ledge- b a s e d DSS can b e larg e, organ ization s h av e inv ested in them . T h e s e D SS are utilized in the creation o f a u to m a te d d ecision -m akin g systems, as d e s c rib e d in C h ap ter 12. T h e b a sic idea is that rules are u se d to au to m ate th e d ecisio n -m ak in g p ro ce ss. T h e s e ru les are b asically e ith er a n ES o r stru ctured like o n e . T h is is im portant w h e n d ecisio n s m ust b e m ad e quickly, as in m an y e -c o m m e r c e situations.

M O D E L-D R IV E N D S S T h e m a jo r e m p h a s e s o f D SS th a t a re prim arily d e v e lo p e d around o n e o r m o re (la rg e -sca le / co m p le x ) op tim izatio n o r sim u latio n m o d e ls ty p ically inclu de sig n ifican t activities in m o d e l fo rm u lation , m o d e l m a in te n a n ce , m o d el m an ag em en t

Chapter 2 • Foundations and T ech n o lo g ies for D ecisio n Making

in distributed co m p u tin g e n v iro n m en ts, a n d w h a t-if analyses. M any la rg e-sc a le a p p lic a ­ tions fall in to this ca te g o ry . N otable e x a m p le s in clu d e th o s e u sed b y P ro c te r & G a m b le (Farasyn e t al., 2 0 0 8 ), H P (O la v s o n an d F iy , 2 0 0 8 ), an d m an y oth ers.

T h e fo c u s o f s u c h system s is o n u sin g th e m o d e l(s ) to o p tim ize o n e o r m o re o b je c ­ tives (e .g ., p ro fit). T h e m o st co m m o n e n d -u se r to o l fo r D SS d ev e lo p m e n t is M icroso ft E xcel. E x c e l in clu d e s d o z e n s o f statistical p a ck a g e s, a lin e a r p rogram m ing p a c k a g e (S o lv er), an d m a n y fin an cial a n d m a n a g e m e n t s c ie n c e m o d e ls. W e w ill study th e s e in m o re d etail in C h ap te r 9- T h e s e D SS ty p ically c a n b e g ro u p e d u n d e r th e n e w la b e l o f prescriptive an alytics.

CO M PO U N D D S S A co m p o u n d , o r h yb rid , D SS in clu d e s tw o o r m o re o f th e m a jo r ca t­ e g o rie s d e s c rib e d e a rlie r. O fte n , a n ES c a n b e n e fit b y u tilizin g s o m e o p tim izatio n , an d clearly a d ata-d riv e n D SS c a n fe e d a la rg e -sc a le o p tim iz a tio n m o d e l. S o m e tim e s d o c u ­ m e n ts a re critical in u n d erstan d in g h o w to in te rp re t th e results o f visu alizin g d ata fro m a d ata-d riven D SS.

A n e m e rg in g e x a m p le o f a c o m p o u n d D SS is a p ro d u ct o ffe re d b y W o lfram A lp h a (w o lfram alp h a.co m ). It c o m p ile s k n o w le d g e fro m o u ts id e d a ta b a s e s , m o d e ls , a lg o ­ rithm s, d o c u m e n ts , a n d s o o n to p ro v id e a n s w e rs to s p e c ific q u e s tio n s . F o r e x a m p le , it c a n fin d an d a n a ly z e c u rre n t d ata fo r a s to c k a n d c o m p a r e it w ith o th e r s to c k s . It c a n a lso te ll y o u h o w m a n y c a lo rie s y o u w ill b u rn w h e n p e rfo rm in g a s p e c ific e x e r c i s e o r th e s id e e ffe c ts o f a p a rticu la r m e d ic in e . A lth o u g h it is in e a rly s ta g e s as a c o lle c tio n o f k n o w le d g e c o m p o n e n ts fro m m a n y d iffe re n t a re a s, it is a g o o d e x a m p le o f a c o m p o u n d D SS in g e ttin g its k n o w le d g e fro m m a n y d iv erse s o u rc e s a n d atte m p tin g t o s y n th e s iz e it.

Other DSS Categories M any o th e r p ro p o s a ls h a v e b e e n m ad e to classify D SS. P erh a p s th e first form al attem p t w a s b y A lter (1 9 8 0 ) . S e v eral o th e r im p o rtan t ca te g o rie s o f D SS in clu d e ( 1 ) in stitu tion al and ad h o c D SS; ( 2 ) p e rso n a l, gro u p , a n d organ izatio n al su p p ort; ( 3 ) individual su p p ort system versu s G SS; a n d ( 4 ) cu sto m -m a d e sy stem s v ersu s re ad y -m ad e sy stem s. W e d iscu ss s o m e o f th e s e n ext.

IN S T IT U T IO N A L A N D A D H O C D S S Institutional DSS (s e e D o n o v a n an d M ad n ick , 1 9 7 7 ) d ea l w ith d ecisio n s o f a recu rrin g natu re. A ty p ical e x a m p le is a p o rtfo lio m a n a g e m e n t system (P M S), w h ic h h a s b e e n u se d b y sev eral larg e b a n k s fo r su p p o rtin g in v e stm e n t d ecisio n s. An institu tionalized D SS c a n b e d e v e lo p e d a n d re fin ed as it e v o lv e s o v e r a n u m b e r o f y e ars, b e c a u s e th e D SS is u s e d rep eated ly to so lv e id en tical o r sim ilar p ro b ­ lem s. It is im p ortan t to re m e m b e r that a n institutional D SS m ay n o t b e u s e d b y e v e ry o n e in a n o rg an ization ; it is th e recu rrin g n a tu re o f th e d ecisio n -m a k in g p ro b lem th a t d eter­ m in es w h e th e r a D SS is institutional v ersu s ad h o c .

Ad h o c DSS d e a l w ith s p e c ific p ro b lem s that a re u su ally n e ith er an ticip ated n o r re cu r­ ring. A d h o c d e cis io n s o fte n in v o lv e strateg ic p lan n in g issu es an d s o m e tim e s m a n a g e m e n t co n tro l p ro b lem s. Ju s tify in g a D SS that w ill b e u s e d o n ly o n c e o r tw ice is a m a jo r issu e in D SS d ev elo p m en t. C o u n tless a d h o c D SS ap p licatio n s h a v e e v o lv e d in to institu tional DSS. E ith er th e p ro b le m recu rs a n d th e sy stem is re u s e d o r oth ers in th e o rg an izatio n have sim ilar n e e d s th a t c a n b e h a n d led b y th e fo rm erly a d h o c D SS.

Custom-Made System s Versus Ready-Made Systems M any D SS a re c u s to m m a d e fo r individ ual u se rs and o rg a n iz a tio n s. H o w e v e r, a c o m ­ p a ra b le p r o b le m m a y e x is t in sim ilar o rg a n iz a tio n s. F o r e x a m p le , h o s p ita ls , b a n k s, a n d u n iv e rs itie s s h a re m a n y sim ilar p ro b le m s . Sim ilarly, c e rta in n o n ro u tin e p ro b le m s in a fu n ctio n a l a re a ( e .g ., fin a n c e , a c c o u n tin g ) c a n re p e a t th e m s e lv e s in t h e s a m e fu n c tio n a l

9 4 Part I • D ec isio n M aking and Analytics: An O verview

a re a o f d iffe re n t a re a s o r o rg a n iz a tio n s. T h e r e fo r e , it m a k e s s e n s e to b u ild g e n e r ic D SS th a t c a n b e u s e d (s o m e tim e s w ith m o d ific a tio n s ) in se v e ra l o rg a n iz a tio n s. S u c h D SS a re c a lle d r e a d y -m a d e an d a re s o ld b y vario u s v e n d o rs (e .g ., C o g n o s , M icroStrategy , T e ra d a ta ). E sse n tia lly , th e d a ta b a s e , m o d e ls, in te rfa c e , a n d o th e r su p p o rt fe a tu re s are b u ilt in: Ju s t a d d a n o r g a n iz a tio n ’s d ata an d lo g o . T h e m a jo r OLAP a n d a n a ly tics v e n d o rs p ro v id e D SS te m p la te s fo r a v arie ty o f fu n c tio n a l a re a s , in clu d in g fin a n c e , re a l e sta te, m ark e tin g , a n d a c c o u n tin g . T h e n u m b e r o f re a d y -m a d e D SS c o n tin u e s to in c re a se b e c a u s e o f th e ir flex ib ility a n d lo w c o s t. T h e y a r e ty p ica lly d e v e lo p e d u sin g In te rn e t te c h n o lo g ie s fo r d a ta b a s e a c c e s s a n d c o m m u n ic a tio n s , and W e b b ro w s e rs fo r in te rfa ce s. T h e y a ls o re a d ily in c o rp o ra te OLAP a n d o th e r e a s y -to -u s e D SS g e n e ra to rs.

O n e c o m p lic a tio n in te rm in o lo g y resu lts w h e n a n o rg a n iz a tio n d e v e lo p s a n in stitu tion al sy stem b u t, b e c a u s e o f its stru ctu re, u s e s it in a n ad h o c m an n er. An o rg an i­ z a tio n c a n b u ild a larg e d ata w a r e h o u s e b u t th e n u s e OLAP to o ls to q u e ry it an d p erfo rm ad h o c an aly sis to s o lv e n o n re cu rrin g p ro b le m s. T h e D SS e x h ib its th e traits o f ad h o c an d institu tional sy stem s a n d a lso o f c u sto m a n d re a d y -m a d e sy stem s. S ev eral ER P, CRM, k n o w le d g e m a n a g e m e n t (K M ), a n d SCM c o m p a n ie s o ffe r D SS a p p lica tio n s o n lin e . T h e s e k in d s o f sy stem s c a n b e v ie w e d as re a d y -m a d e, a lth o u g h typ ically th e y re q u ire m o d ifica­ tio n s (s o m e tim e s m a jo r) b e fo r e th e y c a n b e u s e d effe ctiv e ly .

SECTION 2 .1 0 REVIEW QUESTIONS

1 . List th e D SS classificatio n s o f th e A1S SIG D SS.

2 . D e fin e d ocu m en t-d riv en DSS. 3 . List th e cap ab ilitie s o f institu tional D SS an d a d h o c DSS. 4 . D e fin e th e term rea d y -m a d e DSS.

2.11 C O M PO N EN T S OF D E C IS IO N SU P PO R T S Y S T E M S A D SS a p p lica tio n c a n b e c o m p o s e d o f a d ata m a n a g e m e n t su b sy stem , a m o d e l m an­ a g e m e n t su b sy stem , a user in te rface su b sy stem , an d a k n o w le d g e -b a s e d m an ag em en t su b system . W e sh o w th e s e in Figure 2.4.

FIG U R E 2 .4 Schematic V iew o f DSS.

Chapter 2 • Foundations and T ech n o lo g ies for D ecision M aking 9 5

FIGURE 2.5 Structure of the Data Management Subsystem.

The Data M anagem ent Subsystem T h e data m an a g e m e n t su b sy stem in clu d es a d a ta b a se that co n ta in s re le v a n t d ata fo r th e situ atio n an d is m a n a g e d b y so ftw are c a lle d th e database m an agem en t system (DBMS).2 T h e d ata m a n a g e m e n t su b sy stem c a n b e in te rc o n n e c te d w ith th e co rp o ra te data w areh ou se, a re p o sito ry fo r co rp o ra te re le v a n t d ecisio n -m a k in g d ata. U sually, th e data are s to red o r a c c e s s e d via a d atab ase W e b server. T h e data m a n a g e m e n t su b sy stem is c o m p o s e d o f th e fo llo w in g elem en ts:

• D SS d atab ase • D a ta b a se m a n a g e m e n t system • D ata d irecto ry • Q u e ry facility

T h e s e e le m e n ts a re s h o w n sch e m a tica lly in F ig u re 2.5 (in th e sh a d e d a r e a ). T h e figu re ils o s h o w s th e in te ra ctio n o f th e d ata m a n a g e m e n t su b sy stem w ith th e o t h e r parts o f th e DSS, a s w e ll as its in te ra ctio n w ith se v e ra l d ata s o u rc e s. M any o f th e B l o r d escrip tiv e analytics a p p lica tio n s d eriv e th e ir stren g th fro m th e d ata m a n a g e m e n t sid e o f th e su b sy s­ tem s. A p p licatio n C ase 2 .2 p ro v id e s a n e x a m p le o f a D SS that fo c u s e s o n d ata.

The Model M anagem ent Subsystem T h e m o d e l m a n a g e m e n t su b sy stem is th e c o m p o n e n t th a t in clu d es fin an cial, statistical, m a n a g e m e n t s c ie n c e , o r o th e r quantitative m o d e ls th a t pro v id e th e sy stem ’s analytical cap ab ilitie s a n d ap p ro p riate so ftw are m an ag e m e n t. M o d elin g la n g u a g e s fo r b u ild in g c u s ­ to m m o d e ls a re a lso in clu d ed . T h is so ftw are is o fte n ca lle d a m odel b ase m anagem ent

2DBMS is used as both singular and plural (system and system s), as are many other acronyms in this text.

9 6 Part I • D ec isio n Making and Analytics: An O verview

Application Case 2.2 Station Casinos Wins by Building Customer Relationships Using Its Data Station C asin o s is a m a jo r p ro vid er o f g am in g for Las V e g a s -a r e a resid en ts. It o w n s a b o u t 2 0 p ro p e r­ tie s in N evada a n d o th e r states, em p lo y s o v e r 12 ,0 0 0 p e o p le , a n d h a s re v e n u e o f o v e r $1 b illion.

Station C asin o s w a n te d to d e v e lo p an in -d ep th v ie w o f e a c h cu stom er/ gu est w h o v isited C asino Statio n p ro p e rtie s. T h is w o u ld p erm it th e m to b e t­ te r u n d erstan d cu sto m e r trend s a s w e ll as e n h a n c e th e ir o n e -to -o n e m ark etin g fo r e a c h g u est. T h e c o m ­ p a n y e m p lo y e d th e T erad ata w a re h o u s e to d ev elo p th e “T o ta l G u e s t W o rth ” solu tio n . T h e p ro je c t u se d u se d A p rim o R e latio n sh ip M anager, In fo rm atica, and C o g n o s to ca p tu re , an aly ze, an d s e g m e n t cu stom ers. A lm o st 5 0 0 d iffe ren t data s o u rc e s w e re in teg rated to d e v e lo p th e fu ll v ie w o f a cu sto m er. As a result, the co m p a n y w as a b le to realize th e fo llo w in g b en e fits:

• Custom er seg m en ts w e re e xp an d ed from 14 (originally) to 160 segm ents s o as to b e a b le to target m o re sp ecific p rom otion s to e a ch segm ent.

• A 4 p e rc e n t to 6 p e r c e n t in cre a se in m onthly s lo t profit.

• Slot p ro m otion co sts w e re red u ced b y $1 million (from $13 m illion p e r m onth) b y b etter targeting th e cu stom er segm ents.

• A 14 p e rc e n t im p ro v e m en t in g u e st retention. • In c re a s e d n e w -m e m b e r a cq u isitio n b y 160

p ercen t. • R e d u ctio n in d ata e rro r rates fro m as high as

8 0 p e rc e n t to le s s th a n 1 p ercen t. • R edu ced th e tim e to analyze a cam p aign’s e ffe c­

tiveness fro m alm ost 2 w e e k s to ju st a few hours.

Q u e s t i o n s f o r D i s c u s s i o n

1. W h y is this d e c is io n su p p o rt sy stem classified as a d ata-fo cu sed DSS?

2. W h at w e re s o m e o f th e b e n e fits fro m im p lem en t­ ing this solution?

Source: Teradata.com, “No Limits: Station Casinos Breaks the Mold on Customer Relationships,” teradata.com/case-studies/ S t a t i o n - C a s i n o s - N o - L i m i t s - S t a t i o n - C a s i n o s - B r e a k s - t h e - M o l d - o n -C u s to m e r-R e la tio n s h ip s -E x ec u tiv e -S u m m a ry -e b 6 4 lO (accessed February 2013)-

system (MBMS). T h is c o m p o n e n t c a n b e c o n n e c te d to co rp o ra te o r e x te rn a l storage o f m o d els. M o d el so lu tio n m e th o d s an d m a n a g e m e n t sy stem s a re im p lem e n te d in W e b d ev e lo p m e n t system s (s u c h as Ja v a ) to ru n o n a p p lica tio n serv ers. T h e m o d e l m a n a g e ­ m e n t su b sy stem o f a D SS is c o m p o s e d o f th e fo llo w in g elem en ts;

• M o d el b a se • M BMS • M o d elin g lan g u ag e • M odel d irecto ry • M o d el e x e c u tio n , in teg ratio n , and co m m a n d p ro c e s s o r

T h e s e e le m e n ts and th e ir in te rfa ce s w ith o th e r D SS c o m p o n e n ts are s h o w n in Fig u re 2.6. At a h igh er lev el than build ing b lo ck s, it is im portant to co n sid er the d ifferent typ es o f

m o d els and solu tion m eth o d s n e e d e d in th e DSS. O fte n at th e start o f d ev elop m en t, th e re is so m e s e n se o f th e m o d el typ es to b e incorp o rated , b u t this m ay ch a n g e as m o re is learn ed ab o u t th e d e cisio n p ro blem . So m e D SS d ev elo p m en t system s inclu de a w id e variety o f co m ­ p o n en ts (e .g ., Analytica fro m Lumina D e cisio n System s), w h ere a s others h av e a sin gle o n e (e .g ., Lindo). O ften, th e results o f o n e ty p e o f m o d el c o m p o n e n t (e .g ., fo recastin g ) are u sed as inp u t to an o th er (e .g ., p ro d u ction sched u ling ). In s o m e ca ses, a m o d elin g language is a co m p o n e n t that g e n erates input to a solver, w h ere a s in o th e r ca ses, th e tw o are com bin ed .

B e c a u s e D SS d eal w ith sem istru caired o r un structured p ro b lem s, it is o ften necessary to cu stom ize m o d els, using program m ing to o ls an d languages. So m e e x a m p le s o f th e se are .NET Fram ew ork lang u ages, C++, an d Ja v a . OLAP softw are m ay a lso b e u se d to w o rk w ith m o d els in data analysis. E v e n lan gu ages fo r sim ulation s u ch as A rena and statistical p a ck ­ ages s u ch as th o se o f SPSS o ffe r m o d elin g to o ls d e v e lo p e d through th e u s e o f a proprietary

C hapter 2 • Foundations and T ech n o lo g ies for D ecision Making

FIGURE 2.6 Structure of the Model Management Subsystem.

program m ing lang u age. For sm all an d m ed iu m -sized D SS o r fo r less co m p le x o n e s , a spread­ sheet (e .g ., E x c e l) is usually used. W e w ill u se E x ce l fo r m any k e y e x a m p le s in this b o o k . A pplication C ase 2 .3 d escrib e s a sp re ad sh ee t-b ased DSS. H ow ever, using a sp read sh eet for m o d eling a p ro b le m o f an y significant size p resen ts p ro blem s w ith d o cu m en tatio n and error diagnosis. It is v e iy difficult to d eterm ine o r understand nested , c o m p le x relationships in sp read sh eets crea ted b y s o m e o n e else. This m akes it difficult to m odify a m o d el built b y s o m e o n e e lse . A related issu e is the in cre ase d lik elih o o d o f errors cre e p in g in to th e form u­ las. W ith all th e eq u atio n s ap p earin g in th e fo rm o f cell re fe re n ces, it is ch allen g in g to figure ou t w h ere a n e rro r m ight b e . T h e s e issu es w e re ad dressed in a n early g e n eratio n o f DSS d ev elo p m en t softw are that w as available o n m ainfram e com p u ters in th e 1980s. O n e su ch product w as calle d Interactive Financial P lanning System (IFP S). Its d ev elo p e r. Dr. G erald W agner, th e n re le a se d a d esk to p softw are called P lanners Lab. P lanners Lab includes th e follow ing co m p o n e n ts: (1 ) a n easy -to-u se algeb raically orien ted m od el-bu ild ing language and (2 ) a n e asy -to -u se state-of-the-art o p tio n fo r visualizing m o d el output, s u ch as answ ers to w h at-if a n d g o a l s e e k q u estio ns to analyze results o f ch a n g e s in assum ptions. T h e co m ­ b in ation o f th e s e co m p o n e n ts e n a b le s b u sin ess m anagers and analysts to b u ild , review , and ch allen g e th e assum p tions th at underlie d ecisio n -m akin g scen arios.

P lan n ers L ab m a k e s it p o ss ib le fo r th e d e c is io n m ak ers to “p lay ” w ith assu m p tions to re flect altern ativ e v iew s o f th e fu ture. E v ery P lan n e rs Lab m o d e l is a n a s s e m b la g e o f a ssu m p tio n s a b o u t th e future. A ssu m ptions m ay c o m e fro m d a tab ase s o f h isto rical p e r­ fo rm a n ce , m a rk e t re s e a rch , a n d th e d e c is io n m a k e rs’ m in d s, to n a m e a fe w so u rce s. M ost a ssu m p tio n s a b o u t th e fu ture c o m e fro m th e d e c is io n m a k e rs’ a ccu m u la te d e x p e rie n c e s in th e fo rm o f o p in io n s.

T h e resu ltin g c o lle c tio n o f e q u a tio n s is a P lan n e rs Lab m o d e l th at tells a r e a d a b le story f o r a p a r t ic u la r scen ario . P lan n e rs Lab lets d e c is io n m ak ers d e s c rib e th e ir plans in th eir o w n w o rd s and w ith th e ir o w n assu m p tio n s. T h e p ro d u ct’s ra iso n d ’etre is that a sim ulator sh o u ld facilitate a co n v e rsa tio n w ith th e d e c is io n m a k e r in th e p ro c e s s o f

Part I • D ec isio n M aking and Analytics: An Overview

Application Case 2.3 S N A P DSS Helps O neN et M ak e Telecom m unications R ate Decisions T e le co m m u n ica tio n s n e tw o rk serv ices to ed u cation al institutions a n d g o v ern m e n t entities a re typically pro v id ed b y a m ix o f private and p u b lic organiza­ tions. M any states in th e U n ited States h av e o n e o r m o re state a g e n c ie s that a re re sp o n sib le fo r providing n e tw o rk serv ices to s c h o o ls , co lle g e s, an d o th er state ag e n cie s. O n e e x a m p le o f s u ch a n a g e n c y is O n eN et in O k lah o m a. O n e N e t is a d ivision o f th e O k lah o m a State R eg en ts fo r H ig h er E d u cation an d o p e ra te d in co o p e ra tio n w ith th e O ffice o f State Fin an ce.

U sually a g e n c ie s s u ch as O n e N et o p e ra te as a n e n te rp rise -ty p e fund. T h e y m ust re c o v e r th eir co s ts th ro u g h b illin g th e ir c lie n ts and/or b y justifying ap p ro p riatio n s d irectly fro m th e state legislatu res. T h is c o s t re c o v e ry sh o u ld o c c u r th ro u g h a p ricing m e ch a n ism th a t is e fficie n t, sim p le to im p lem en t, a n d e q u ita b le. T h is p ricing m o d e l ty p ically n e e d s to re co g n iz e m an y facto rs: c o n v e rg e n c e o f v o ic e , data, a n d v id eo traffic o n th e sa m e infrastructure; diver­ sity o f u s e r b a s e in term s o f e d u ca tio n a l institutions, sta te a g e n c ie s , a n d s o o n ; diversity o f ap p licatio n s in u s e b y state clie n ts, fro m e-m ail to v id e o c o n fe r­ e n c e s , IP te le p h o n in g , an d d istan ce learn in g; re co v ­ ery o f cu rre n t c o s ts , a s w ell a s p lan n in g fo r up grad es

a n d fu ture d ev elo p m en ts; a n d lev erag e o f th e sh ared infrastru ctu re to e n a b le fu rth er e c o n o m ic d e v e lo p ­ m e n t a n d co lla b o ra tiv e w o rk a cro ss th e state that lead s to in n ov ativ e u s e s o f O neN et.

T h e s e c o n sid e ra tio n s le d to th e d e v e lo p m e n t o f a sp re a d sh e e t-b a s e d m o d e l. T h e system , SNAP-DSS, o r Serv ice N etw o rk A p p licatio n a n d P ricin g (SNAP)- b ase d D SS, w a s d e v e lo p e d in M icroso ft E x c e l 2 0 0 7 an d u se d th e V BA p ro g ram m in g langu age.

T h e SNAP-DSS o ffers O n e N et th e ability to s ele ct th e rate card op tio n s th a t b e s t fit th e preferred p ric­ ing strategies b y providing a real-tim e, user-friendly, graphical u ser in terface (G U I). In addition, th e SNAP- D SS n o t o n ly illustrates th e in flu ence o f the ch an g e s in th e p ricing factors o n e a c h rate card o p tio n , but also allow s th e u ser to an alyze various rate card op tio ns in d ifferent scen ario s u sin g different param eters. This m o d el has b e e n u sed b y O n eN et financial p lanners to gain insights into th e ir cu stom ers an d analyze m any w h at-if s cen ario s o f d ifferent rate p lan op tions.

Source: Based on J. Chongwatpol and R. Sharda, “SNAP: A DSS to Analyze Network Service Pricing for State Networks,” D ecision Support Systems, Vol. 50, No. 1, December 2010, pp. 347-359.

d escrib in g b u sin e ss assu m p tio ns. All assu m p tio n s are d e s c rib e d in English e q u a tio n s (o r th e u ser’s native lan g u ag e).

T h e b e s t w a y to learn h o w to use P lan n ers L ab is to la u n ch th e so ftw are an d fo llo w th e tutorials. T h e softw are ca n b e d o w n lo a d e d at p l a n n e r s l a b . c o m .

The User Interface Subsystem T h e u s e r c o m m u n ic a te s w ith a n d co m m a n d s th e D S S th ro u g h th e u s e r i n t e r f a c e s u b ­ sy stem . T h e u s e r is c o n s id e re d p art o f th e sy stem . R e s e a rc h e rs a ss e rt th a t s o m e o f th e u n iq u e c o n trib u tio n s o f D SS a re d eriv ed fro m th e in te n siv e in te ra c tio n b e tw e e n th e c o m p u te r an d th e d e c is io n m ak e r. T h e W e b b r o w s e r p ro v id e s a fam iliar, co n s is te n t g ra p h ica l u s e r in te rfa c e (G U I) stru ctu re fo r m o st D SS. F o r lo c a lly u s e d D SS, a s p re a d ­ s h e e t a ls o p ro v id e s a fam iliar u s e r in te rfa c e . A d ifficu lt u s e r in te rfa ce is o n e o f th e m a jo r re a s o n s m a n a g e rs d o n o t u s e c o m p u te rs a n d q u a n tita tiv e a n a ly se s a s m u c h as th e y c o u ld , g iv e n th e a v ailab ility o f th e s e te c h n o lo g ie s . T h e W e b b r o w s e r h a s b e e n re c o g n iz e d a s a n e ffe c tiv e D SS G U I b e c a u s e it is fle x ib le , u s e r frien d ly , a n d a g a te w a y to a lm o st all s o u r c e s o f n e c e s s a ry in fo rm a tio n a n d d ata. E sse n tia lly , W e b b ro w s e rs h a v e le d to th e d e v e lo p m e n t o f p o rtals and d a s h b o a rd s , w h ic h fro n t e n d m a n y D SS.

E xp losive gro w th in p o rtab le d ev ices inclu ding sm artp h on es an d tablets has ch an g e d th e D SS u ser interfaces as w ell. T h e s e d ev ices allow e ith er h and w ritten input o r typed input fro m internal o r e xte rn al k ey board s. So m e D SS u ser in te rface s utilize natural-language input

Chapter 2 • Foundations and T ech n o lo g ies fo r D ecision Making 9 9

- r .. te x t in a h u m an lan g u ag e) s o that th e users c a n easily e x p ress th e m se lv es in a m ean - mgful way. B e c a u s e o f th e fuzzy natu re o f hu m an langu age, it is fairly difficult to d ev elop software to interpret it. H ow ever, th e se p ack ag e s in cre ase in a ccu racy e v e ry y ear, and they will ultim ately lead to accu rate input, output, and language translators.

C ell p h o n e inputs th ro u g h SMS are b e c o m in g m o re co m m o n fo r a t le a s t s o m e c o n ­ su m er D SS-ty p e ap p licatio n s. F o r e x a m p le , o n e c a n s en d a n SMS re q u e s t fo r s e a rch on any to p ic to G O O G L (4 6 6 4 5 ) . It is m o st u sefu l in lo catin g n e a rb y b u s in e s s e s , ad d resses, o r p h o n e n u m b e rs, b u t it ca n a lso b e u se d fo r m an y o th e r d e c is io n su p p o rt tasks. F or e xam p le, u se rs c a n fin d d efin ition s o f w o rd s b y e n te rin g th e w o rd “d e fin e ” fo llo w e d b y a w ord, s u ch a s “d efin e e x te n u a te .” S o m e o f th e o th e r cap ab ilitie s in clu d e:

• T ran slatio n s: “T ra n sla te th a n k s in S p a n ish .” • P rice lo o k u p s : “P rice 3 2 G B iP h o n e .” • C alcu lator: A lthough y o u w o u ld p ro b a b ly ju st w an t to u s e y o u r p h o n e ’s built-in

c a lcu la to r fu n ctio n , y o u c a n s en d a m ath e x p r e s s io n as a n SMS fo r a n answ er. • C u rren cy co n v e rsio n s: “10 u sd in e u r o s .” • Sports s co re s and gam e times: Ju s t en ter the nam e o f a team ( “NYC G iants”), and G oo g le

SMS w ill sen d the m ost re ce n t g am e’s sco re an d th e date and tim e o f th e next match.

This typ e o f SM S-based s e a rch cap ab ility is a lso av ailab le fo r o th e r s e a r c h e n g in e s , inclu d ­ ing Y a h o o ! a n d M icroso ft’s n e w s e a rch e n g in e B in g .

W ith th e e m e r g e n c e o f s m a rtp h o n e s s u c h as A p p le ’s iP h o n e a n d A n d ro id sm art­ p h o n e s