week 8-data science
rav
Chapter_121.pptxData Science and Big Data Analytics
Chapter 12: The Endgame, or Putting It All Together
1
Chapter Contents
12.1 Communicating and Operationalizing an Analytics Project
12.2 Creating the Final Deliverables
Developing core material for multiple audiences, project goals, main findings, approach, model description, key points supported with data, model details, recommendations, tips on final presentation, providing technical specifications and code
12.3 Data Visualization Basics
Key points supported with data, evolution of a graph, common representation methods, how to clean up a graphic, additional considerations
Summary
2
12.1 Communicating and Operationalizing an Analytics Project
3
12.1 Communicating and Operationalizing an Analytics Project Deliverables and Stakeholders
4
12.1 Communicating and Operationalizing an Analytics Project Deliverables
General Deliverables – from Textbook
Presentation for Project Sponsors
Presentation for Analysts
Code
Technical Specifications
Deliverables For This Course
Presentation for Analysts – half hour per team, next week
Technical Paper for Research Day Conference
Submit CD – Presentation, Paper, Data or URL, Code
5
12.2 Creating the Final Deliverables Case Study – Fictional Bank Churn Prediction
This section describes a scenario of a fictional bank and a churn prediction model of its customers
The analytic plan contains components that can be used as inputs for writing the final presentations
scope
underlying assumptions
modeling techniques
initial hypotheses
and key findings
6
12.2 Creating the Final Deliverables Case Study – Fictional Bank Churn Prediction
7
12.2 Creating the Final Deliverables Case Study – Fictional Bank Analytics Plan
8
12.2 Creating the Final Deliverables 12.2.1 Developing Core Material for Multiple Audiences
Some project components have dual use
Create core materials used for both analyst and business audiences
Three areas on the next slide used for both audiences
Sections after the following overview slide
12.2.2 – Project Goals
12.2.3 – Key Findings
12.2.4 – Approach
12.2.5 – Model Description
12.2.6 – Key Points Supported by Data
12.2.7 – Model Details
12.2.8 – Recommendations
12.2.9 – Additional Tips on the Final Presentation
12.2.10 – Providing Technical Specifications and Code
9
12.2 Creating the Final Deliverables 12.2.1 Developing Core Material for Multiple Audiences
10
12.2 Creating the Final Deliverables 12.2.2 Project Goals
The project goals portion of the final presentation is generally the same for sponsors and analysts
The project goals are described first to lay the groundwork for the solution and recommendations
Generally, the goals are agreed on early in the project
Two examples of project goals are shown next
The second example recaps the situation that motivated the project
11
12.2 Creating the Final Deliverables 12.2.2 Project Goals
12
12.2 Creating the Final Deliverables 12.2.2 Project Goals
13
12.2 Creating the Final Deliverables 12.2.3 Main Findings
14
12.2 Creating the Final Deliverables 12.2.3 Main Findings
Sponsor Service Level Agreement
15
12.2 Creating the Final Deliverables 12.2.4 Approach
16
12.2 Creating the Final Deliverables 12.2.4 Approach
17
12.2 Creating the Final Deliverables 12.2.5 Model Description
18
12.2 Creating the Final Deliverables 12.2.6 Key Points Supported with Data
Identify key points based on insights and observations from the data and model results
This information lays the foundation for the coming recommendations
19
12.2 Creating the Final Deliverables 12.2.6 Key Points Supported with Data
Rate of bank customers who would churn
20
12.2 Creating the Final Deliverables 12.2.7 Model Details
21
12.2 Creating the Final Deliverables 12.2.7 Model Details
Caption: Model details comparing two data variables
22
12.2 Creating the Final Deliverables 12.2.8 Recommendations
23
12.2 Creating the Final Deliverables 12.2.9 Additional Tips on Final Presentation
Use imagery and visual representations
Pictures are better than words
Ensure text mutually exclusive/collectively exhaustive
Meaning: cover key points, but don’t repeat unnecessarily
Measure and quantify the benefits of the project
Requires effort to do this well
Make the project benefits clear and conspicuous
Details
Provide sufficient context for recommendations
Spell out acronyms, avoid excessive jargon
24
12.2 Creating the Final Deliverables 12.2.10 Providing Technical Specifications and Code
Deliver code plus documentation
User manual
Add extensive comments in the code
How computationally expensive to run the model?
Describe exception handling
Data outside expected data ranges
Null values
Zeros
25
12.3 Data Visualization Basics Common Tools for Data Visualization
26
12.3 Data Visualization Basics 12.3.1 Key Points Supported with Data
Difficult to make key insights when data is in tables
Text shows 45 then 35 years of store operations
Ten years shown here
27
12.3 Data Visualization Basics 12.3.1 Key Points Supported with Data
Shows where the BigBox store has market saturation
28
12.3 Data Visualization Basics 12.3.2 Evolution of a Graph
Visualization can allow people to understand data on an intuitive, precognitive level
Distribution of customer (user) loyalty scores
Log scale
Less skewed
29
12.3 Data Visualization Basics 12.3.2 Evolution of a Graph
Rescaled view of last figure with median ~ 2.0
Textbook does not describe the rescaling method
30
12.3 Data Visualization Basics 12.3.2 Evolution of a Graph
Graph of stability analysis (over time) for pricing
31
12.3 Data Visualization Basics 12.3.2 Evolution of a Graph
Current pricing model with score distribution (rug)
32
12.3 Data Visualization Basics 12.3.2 Evolution of a Graph
Proposed pricing model with loyalty score dist. (rug)
Proposes progressively higher prices as customer loyalty increases
May seem counterintuitive
33
12.3 Data Visualization Basics 12.3.2 Evolution of a Graph
Evolution of a Graph, Analyst Example
34
12.3 Data Visualization Basics 12.3.2 Evolution of a Graph
Evolution of a Graph, Sponsor Example
35
12.3 Data Visualization Basics 12.3.3 Common Representation Methods
36
12.3 Data Visualization Basics 12.3.4 How to Clean Up a Graphic
Example 1
Before
37
12.3 Data Visualization Basics 12.3.4 How to Clean Up a Graphic
Example 1
After
38
12.3 Data Visualization Basics 12.3.4 How to Clean Up a Graphic
Example 2
Before
39
12.3 Data Visualization Basics 12.3.4 How to Clean Up a Graphic
Example 2
After
40
12.3 Data Visualization Basics 12.3.4 How to Clean Up a Graphic
Example 2
After
Alternative
41
12.3 Data Visualization Basics 12.3.5 Additional Considerations
Simple bar chart with two dimensions
42
12.3 Data Visualization Basics 12.3.5 Additional Considerations
Avoid three dimensions
Distort scales and axes, impede viewer cognition
43
Summary
Communicating the value of analytical projects is critical for sustaining the momentum of a project and building support within organizations
Deliverables to satisfy various stakeholders
Presentation for project sponsor
Presentation for analytical audience
Technical specification documents
Well-annotated production code
Best data visualizations are simple and clear
44
