Inhaltsverzeichnis

Alle Kapitel aufklappen
Alle Kapitel zuklappen
Preface
15
Objective of This Book
15
Target Audience
16
Structure of This Book
16
Part I: Getting Started
17
Part II: Time Series Forecasting Models
17
Part III: Classification Models and Regression Models
18
Acknowledgments
18
Conclusion
19
I Getting Started
21
1 An Introduction to Predictive Analytics in SAP Analytics Cloud
23
1.1 The Importance of Predictive Analytics
23
1.1.1 A Brief History of Predictive Analytics
23
1.1.2 The Role of Predictive Analytics in Business
24
1.2 Predictive Analytics in SAP Analytics Cloud
26
1.2.1 A Brief History of Predictive Analytics in SAP Analytics Cloud
26
1.2.2 Business Intelligence, Enterprise Planning, and Augmented Analytics
27
1.3 Customer Use Cases
30
1.3.1 Time Series Forecasting
30
1.3.2 Classification and Regression
33
1.4 Summary
34
2 What Are Predictive Scenarios?
35
2.1 Introducing Predictive Scenarios
35
2.1.1 Using and Securing Predictive Scenarios
35
2.1.2 User Interface of Predictive Scenarios
37
2.2 The Different Types of Predictive Scenarios
41
2.3 The Predictive Ecosystem
42
2.3.1 Datasets
42
2.3.2 Planning Models
44
2.3.3 Stories
48
2.3.4 Multi Actions
49
2.3.5 Data Actions
50
2.4 Summary
51
3 Predictive Analytics Projects
53
3.1 Predictive Analytics Project Stakeholders
53
3.1.1 Business Owners
53
3.1.2 Data Engineers
54
3.1.3 Prediction Creators
54
3.1.4 Story Designers
55
3.1.5 Prediction Consumers
56
3.1.6 Information Technology
56
3.2 How to Implement a Predictive Analytics Project
56
3.2.1 Business Understanding
58
3.2.2 Data Understanding
59
3.2.3 Data Preparation
62
3.2.4 Predictive Modeling
64
3.2.5 Predictive Model Evaluation
65
3.2.6 Delivering Predictions
67
3.3 Summary
68
II Time Series Forecasting Models
69
4 Introducing Time Series Forecasting Models
71
4.1 What Is Time Series Forecasting?
71
4.1.1 Key Terms and Concepts
71
4.1.2 Top-Down versus Bottom-Up Predictive Forecasting
74
4.2 Data Sources for Time Series Forecasting Models
76
4.2.1 Data Sources
76
4.2.2 Basic Data Preparation for Datasets
77
4.3 End-to-End Time Series Forecasting Workflows
82
4.3.1 Time Series Forecasting Based on Planning Models
82
4.3.2 Time Series Forecasting Based on Datasets
83
4.4 Summary
84
5 Using Predictive Forecasts in the Planning Process
85
5.1 Business Scenario
85
5.2 Planning Models
86
5.2.1 Measures
87
5.2.2 Dimensions
87
5.2.3 Versions
91
5.2.4 Currencies
92
5.3 Creating Time Series Forecasting Models with Predictive Planning
92
5.3.1 Create a Predictive Scenario
93
5.3.2 General Settings
94
5.3.3 Predictive Goal
95
5.3.4 Predictive Model Training
99
5.3.5 Influencers
102
5.3.6 Entities
103
5.4 Understanding Time Series Forecasting Models
108
5.4.1 Assessing Model Performance
109
5.4.2 Time Series Forecasting Model Reports
114
5.5 Improving Time Series Forecasting Models with Predictive Planning
120
5.5.1 Finding the Optimal Training Dataset Size
121
5.5.2 Filtering Entities
122
5.5.3 Adding Influencers
123
5.6 Saving Predictive Forecasts
124
5.6.1 Creating a Private Version
124
5.6.2 Saving Forecasts into a Private Version
125
5.7 Including Predictive Forecasts in a Story
126
5.7.1 Including Predictive Forecasts in a Table
126
5.7.2 Including Predictive Forecasts in a Time Series Chart
131
5.8 Summary
132
6 Automating the Production of Predictive Forecasts
133
6.1 Introducing Multi Actions
133
6.2 Creating Predictive Steps in Multi Actions
134
6.3 Using Multi Actions
138
6.3.1 Scheduling Multi Actions
138
6.3.2 Triggering Multi Actions
140
6.4 Summary
142
7 Time Series Forecasting Models Using Datasets
143
7.1 Business Scenario
143
7.2 Creating and Editing Datasets
144
7.2.1 Creating the Training Dataset
144
7.2.2 Additional Dataset Considerations
146
7.3 Creating Time Series Forecasting Models on Datasets
147
7.3.1 General Settings
149
7.3.2 Predictive Goal
150
7.3.3 Predictive Model Training
151
7.3.4 Datasets versus Planning Models
152
7.4 Understanding Time Series Forecasting Models Based on Datasets
154
7.4.1 Assessing Model Performance
154
7.4.2 Time Series Forecasting Model Reports
155
7.5 Improving Time Series Forecasting Models Based on Datasets
157
7.5.1 Finding the Optimal Training Data Size
158
7.5.2 Filtering Entities
158
7.5.3 Adding Influencers
158
7.6 Saving Predictive Forecasts
158
7.7 Including Predictive Forecasts in a Story
160
7.7.1 Report on Predictive Forecasts in a Table
162
7.7.2 Report on Predictive Forecasts in a Time Series Chart
164
7.8 Summary
167
8 Best Practices and Tips for Time Series Forecasting Models
169
8.1 Going Beyond 1,000 Entities
169
8.1.1 Forecast by Batch of 1,000 Entities
170
8.1.2 Aggregate to Predict and Disaggregate to Plan and Report
171
8.1.3 Compared Approaches
172
8.2 Handling Time Series with Missing Data
173
8.3 Considering Time Granularities
177
8.3.1 Using Datasets
177
8.3.2 Using Planning Models
179
8.3.3 Additional Recommendations
179
8.4 Creating Ad Hoc Performance Indicators
180
8.4.1 Saving Forecasts for Past Periods
180
8.4.2 Custom Performance Indicators
181
8.5 Generating What-If Simulations
182
8.6 Forecasting Data at the Right Level
184
8.6.1 Maximizing Forecasting Accuracy
184
8.6.2 Data Quality Is Key
184
8.6.3 Predictive Planning versus Planning
185
8.6.4 Scalability
185
8.6.5 Key Takeaways
186
8.7 Summary
186
9 The Data Science behind Time Series Forecasting Models
187
9.1 End-to-End Process
187
9.2 Additive Modeling Technique
188
9.2.1 Trend
189
9.2.2 Cycles
192
9.2.3 Influencers
193
9.2.4 Fluctuations
193
9.3 Exponential Smoothing
194
9.3.1 Simple Exponential Smoothing
194
9.3.2 Double Exponential Smoothing
195
9.3.3 Triple Exponential Smoothing
197
9.3.4 Damped Trend
198
9.4 Predictive Model Selection
199
9.5 Summary
200
III Classification Models and Regression Models
201
10 Introducing Classification Models and Regression Models
203
10.1 What Is Classification?
203
10.1.1 Key Terms and Concepts
203
10.1.2 Binary Classification
205
10.1.3 Predicted Category and Probability
206
10.2 What Is Regression?
207
10.2.1 Key Terms and Concepts
207
10.2.2 Predictions
208
10.3 Data Sources for Classification Models and Regression Models
209
10.3.1 Data Sources
209
10.3.2 Basic Data Preparation for Datasets
210
10.4 End-to-End Workflow
213
10.5 Summary
214
11 Creating Classification Insights to Enrich Stories
217
11.1 Business Scenario
217
11.2 Using Datasets with Classification Models
218
11.2.1 Training Dataset
218
11.2.2 Application Dataset
221
11.2.3 Predictions Dataset
222
11.3 Creating Classification Models
222
11.3.1 Create a Predictive Scenario
223
11.3.2 General Settings
223
11.3.3 Predictive Goal
226
11.3.4 Influencers
226
11.4 Understanding Classification Models
228
11.4.1 Assessing Model Performance
228
11.4.2 Influencer Contributions
236
11.4.3 Confusion Matrix
240
11.4.4 Profit Simulation
244
11.5 Improving Classification Models
245
11.5.1 Basic Checks
245
11.5.2 Improving Model Accuracy
246
11.5.3 Improving Model Robustness
247
11.5.4 The Quality versus Robustness Trade-Off
248
11.6 Applying Classification Models
248
11.7 Enriching Stories with Classification Insights
251
11.7.1 Prediction Probability
252
11.7.2 Prediction Explanations
255
11.8 Summary
260
12 Creating Regression Insights to Enrich Stories
261
12.1 Business Scenario
261
12.2 Using Datasets with Regression Models
262
12.2.1 Training Dataset
262
12.2.2 Application Dataset
265
12.2.3 Predictions Dataset
265
12.3 Creating Regression Models
266
12.3.1 Create a Predictive Scenario
266
12.3.2 General Settings
267
12.3.3 Predictive Goal
268
12.3.4 Influencers
269
12.4 Understanding Regression Models
270
12.4.1 Assessing Model Performance
270
12.4.2 Influencer Contributions
273
12.5 Improving Regression Models
275
12.5.1 Improving Model Accuracy
275
12.5.2 Improving Model Robustness
275
12.5.3 The Quality versus Robustness Trade-Off
276
12.6 Applying Regression Models
276
12.7 Enriching Stories with Regression Insights
279
12.7.1 Predicted Value
279
12.7.2 Prediction Explanations
281
12.8 Summary
285
13 The Data Science behind Classification Models and Regression Models
287
13.1 Fitting a Predictive Function
287
13.2 Prerequisites to Generating a Model
290
13.3 Evaluating the Model Performance
291
13.3.1 Performance Indicators for Classification Models
291
13.3.2 Performance Indicators for Regression Models
295
13.3.3 Evaluating the Future Performance of the Model
296
13.4 End-to-End Automated Modeling
297
13.4.1 Automated Data Encoding
297
13.4.2 Data Partitions
298
13.4.3 Gradient Boosting
298
13.4.4 Model Performance Evaluation
301
13.5 Generating Predictive Insights
302
13.5.1 Predicted Category
302
13.5.2 Prediction Explanations
304
13.6 Summary
305
14 Conclusion
307
14.1 Lessons Learned
307
14.2 The Future of Predictive Scenarios in SAP Analytics Cloud
307
14.3 Your Next Steps
308
The Authors
309
Index
311