Do on 4/23/2020 2:20PM 20-25 questions Multiple choice timed Chapter 11 Predictive Machine Learning and Chapter 12 Analytics in Practice

4/15/2020

1

Practical Analytics Chapter 12: Analytics in Practice

Timo Elliott © 2015)

Outline

THE DECISION CYCLE

FEEDBACK LOOP AND

OPTIMIZATION

RESPONSIBILITIES OF THE ANALYST

AUTOMATING DECISION MAKING

EXAMPLES OF THE DECISION

CYCLE

SUMMARY

4/15/2020

2

Analytics Techniques

Overview of Analytics Techniques

Exploration and Reporting

• Slicing/dicing

• Multidimensional analysis

• Reporting

Visualization

• Charts

• Dashboards

Knowledge discovery

• Forecasting

• Unsupervised machine learning

• Predictive machine learning

D a

ta S

ta g

in g

P u

b li

sh in

g The Decision Cycle

Data analytics is a component of the cycle that produces actionable decisions and evaluation of results.

The goal of the decision cycle is to use data to make decisions that lead to desired outcomes.

4/15/2020

3

Data

Acquiring and staging data,

covered in Chapters 2,3,4

Data quality is very important

Singe source of truth

Analysis

Covered in Chapters 5 – 12 ◦Slicing and dicing ◦Data visualization ◦Reports and Dashboards ◦Data mining and Big Data ◦Machine Learning ◦Descriptive ◦Predictive

◦Forecasting

4/15/2020

4

Insight(s)

BUSINESS SCENARIO +

ANALYSIS

INTERPRET RESULTS AND DERIVE

INSIGHTS

EXPERIENCE AND DOMAIN

FAMILIARITY

COVERED IN CHAPTERS 6, 9, 10,

11

Decision

Strategists, decision makers,

managers, C-suite

Data-driven decision making is

becoming common

4/15/2020

5

Action

One or more actions require participation of

several divisions and managers

Implement data-driven decisions

Can have short term or long term business impact

Outcome

Results of actions are collected – data

Metrics, KPIs, measures and other ways to quantify outcome

4/15/2020

6

Assessment

Metrics are compared to the desired goals or previous outcomes

Has the decision cycle led to desired improvements?

Use of balanced scorecards

Improvement • If assessments indicated that the outcome of

actions has fallen short of goals → improvements

• Improvements to the decision cycle

• Or improvements to the business functions

• Check to see if analysis is flawed

• Repeat the decision cycle

4/15/2020

7

Feedback Loop and Optimization

A feedback loop is a control mechanism that is employed in economics, the sciences, and engineering to bring actual outcomes into alignment with desired outcomes.

◦ Positive feedback - amplification

◦ Negative feedback – mitigation

Challenges to Optimization

4/15/2020

8

Responsibilities of the Analyst

Analysis-paralysis and validation

Beware of overfitting model to data

Biases in Analytics

• In data collection phase

• In analysis phase

• In insight phase

• In outcome phase

• In assessment phase

• In improvements phase

Admitting analyst biases

BIAS

HTTPS://WWW.EXPLAINXKCD.COM/WIKI/INDEX.PHP/2048:_CURVE-FITTING

4/15/2020

9

Automating Decision Making

Fully manual decision cycle – humans at every phase

Partly automated decision cycle – human/computer combination

Fully automated decision cycle – once trained, computers automate the decision process

• Expert systems

• Artificial intelligence

• Machine learning

Examples of the Decision Cycle

Airline industry pricing

Netflix recommendation

engine

Baseball and the Oakland As

Ford motors and sustainable product

design

4/15/2020

10

Summary

The phases of the cycle are data acquisition, analysis, insight, decision, action, outcome, assessment, and improvement.

The cycle is continuous and uses feedback loops to assist in the optimization of results to goals.

The analytics cycle can be manual, partially automated, or fully automated depending

Challenges to the data-driven decision cycle and to the analysts to keep the analysis, decisions, and actions as free of bias as possible.

Analysts have a responsibility to validate results and to help eliminate biases.

Examples of analytics in decision cycles