Data Mining for Business Analytics: Concepts, Techniques, and Applications with XLMiner

Data Mining For Business Analytics: Concepts, Techniques, and Applicationswith XLMiner®......Page 1
Contents......Page 11
Foreword......Page 23
Preface to the third edition......Page 25
Preface to the first edition......Page 29
Acknowledgments......Page 31
Part I: Preliminaries......Page 35
1.1 What Is Business Analytics?......Page 37
1.3 Data Mining and Related Terms......Page 39
1.4 Big Data......Page 41
1.6 Why Are There So Many Different Methods?......Page 42
1.7 Terminology and Notation......Page 43
Order of Topics......Page 45
2.1 Introduction......Page 49
Prediction......Page 50
Data Reduction and Dimension Reduction......Page 51
Supervised and Unsupervised Learning......Page 52
2.3 The Steps in Data Mining......Page 53
Sampling from a Database......Page 55
Oversampling Rare Events in Classification Tasks......Page 56
Preprocessing and Cleaning the Data......Page 57
2.5 Predictive Power and Overfitting......Page 61
Creation and Use of Data Partitions......Page 62
Overfitting......Page 64
Predicting Home Values in the West Roxbury Neighborhood......Page 66
Modeling Process......Page 68
2.7 Using Excel for Data Mining......Page 75
2.8 Automating Data Mining Solutions......Page 76
Data Mining Software Tools: the State of the Market (by Herb Edelstein)......Page 77
Problems......Page 81
Part II: Data Exploration and Dimension Reduction......Page 84
3.1 Uses of Data Visualization......Page 86
Example 1: Boston Housing Data......Page 88
3.3 Basic Charts: Bar Charts, Line Graphs, and Scatter Plots......Page 89
Distribution Plots: Boxplots and Histograms......Page 91
Heatmaps: Visualizing Correlations and Missing Values......Page 94
Adding Variables: Color, Size, Shape, Multiple Panels, and Animation......Page 96
Manipulations: Re-scaling, Aggregation and Hierarchies, Zooming, Filtering......Page 98
Scaling up to Large Datasets......Page 101
Multivariate Plot: Parallel Coordinates Plot......Page 103
Interactive Visualization......Page 104
Visualizing Networked Data......Page 107
Visualizing Hierarchical Data: Treemaps......Page 109
Visualizing Geographical Data: Map Charts......Page 110
Time Series Forecasting......Page 112
Unsupervised Learning......Page 113
Problems......Page 114
4.1 Introduction......Page 116
4.3 Practical Considerations......Page 117
4.4 Data Summaries......Page 118
Summary Statistics......Page 120
Pivot Tables......Page 121
4.5 Correlation Analysis......Page 122
4.6 Reducing the Number of Categories in Categorical Variables......Page 123
4.8 Principal Components Analysis......Page 124
Example 2: Breakfast Cereals......Page 125
Principal Components......Page 129
Normalizing the Data......Page 131
4.9 Dimension Reduction Using Regression Models......Page 134
4.10 Dimension Reduction Using Classification and Regression Trees......Page 135
Problems......Page 136
Part III: Performance Evaluation......Page 138
Chapter 5: Evaluating Predictive Performance......Page 140
5.2 Evaluating Predictive Performance......Page 141
Prediction Accuracy Measures......Page 142
Comparing Training and Validation Performance......Page 143
Lift Chart......Page 144
Class Separation......Page 146
The Classification Matrix......Page 147
Using the Validation Data......Page 148
Propensities and Cutoff for Classification......Page 149
Performance in Unequal Importance of Classes......Page 153
Asymmetric Misclassification Costs......Page 155
5.4 Judging Ranking Performance......Page 158
Lift Charts for Binary Data......Page 159
Beyond Two Classes......Page 161
Lift Charts Incorporating Costs and Benefits......Page 162
5.5 Oversampling......Page 163
Evaluating Model Performance If Only Oversampled Validation Set Exists......Page 166
Problems......Page 169
Part IV: Prediction and Classification Methods......Page 172
6.1 Introduction......Page 174
6.2 Explanatory vs. Predictive Modeling......Page 175
6.3 Estimating the Regression Equation and Prediction......Page 177
Example: Predicting the Price of Used Toyota Corolla Cars......Page 178
Reducing the Number of Predictors......Page 181
How to Reduce the Number of Predictors......Page 182
Problems......Page 187
7.1 The k-NN Classifier (categorical outcome)......Page 191
Classification Rule......Page 192
Example: Riding Mowers......Page 193
Choosing k......Page 194
Setting the Cutoff Value......Page 195
k-NN with More Than Two Classes......Page 196
7.2 k-NN for a Numerical Response......Page 197
7.3 Advantages and Shortcomings of k -NN Algorithms......Page 199
Problems......Page 201
8.1 Introduction......Page 203
Example 1: Predicting Fraudulent Financial Reporting......Page 204
8.2 Applying the Full (Exact) Bayesian Classifier......Page 205
Practical Difficulty with the Complete (Exact) Bayes Procedure......Page 206
Solution: Naive Bayes......Page 207
Example 2: Predicting Fraudulent Financial Reports, Two Predictors......Page 209
Example 3: Predicting Delayed Flights......Page 210
8.3 Advantages and Shortcomings of the Naive Bayes Classifier......Page 214
Problems......Page 218
Chapter 9: Classification and Regression Trees......Page 220
9.1 Introduction......Page 221
Recursive Partitioning......Page 222
Example 1: Riding Mowers......Page 223
Measures of Impurity......Page 225
Classifying a New Observation......Page 229
Example 2: Acceptance of Personal Loan......Page 230
Stopping Tree Growth: CHAID......Page 233
Pruning the Tree......Page 235
9.5 Classification Rules from Trees......Page 240
Prediction......Page 241
Evaluating Performance......Page 242
9.8 Advantages, Weaknesses and Extensions......Page 243
9.9 Improving Prediction: Multiple Trees......Page 244
Problems......Page 248
Chapter 10: Logistic Regression......Page 252
10.1 Introduction......Page 253
10.2 The Logistic Regression Model......Page 254
Example: Acceptance of Personal Loan......Page 256
Model with a Single Predictor......Page 257
Estimating the Logistic Model from Data: Computing Parameter Estimates......Page 259
Interpreting Results in Terms of Odds (for a Profiling Goal)......Page 261
10.3 Evaluating Classification Performance......Page 263
10.4 Example of Complete Analysis: Predicting Delayed Flights......Page 265
Model Fitting and Estimation......Page 268
Model Interpretation......Page 269
Model Performance......Page 270
Variable Selection......Page 271
Appendix A: Why Linear Regression Is Problematic for a Categorical Response......Page 274
Appendix B: Evaluating Explanatory Power......Page 275
Appendix C: Logistic Regression for More Than Two Classes......Page 278
Problems......Page 281
11.1 Introduction......Page 284
11.2 Concept and Structure of a Neural Network......Page 285
Example 1: Tiny Dataset......Page 286
Computing Output of Nodes......Page 287
Preprocessing the Data......Page 290
Training the Model......Page 291
Example 2: Classifying Accident Severity......Page 295
Avoiding Overfitting......Page 297
Using the Output for Prediction and Classification......Page 298
11.4 Required User Input......Page 300
11.6 Advantages and Weaknesses of Neural Networks......Page 302
Unsupervised Feature Extraction and Deep Learning......Page 304
Problems......Page 305
12.1 Introduction......Page 307
Example 1: Riding Mowers......Page 308
12.2 Distance of an Observation from a Class......Page 309
12.3 Fishers Linear Classification Functions......Page 312
12.4 Classification Performance of Discriminant Analysis......Page 315
12.5 Prior Probabilities......Page 316
12.6 Unequal Misclassification Costs......Page 317
Example 3: Medical Dispatch to Accident Scenes......Page 318
12.8 Advantages and Weaknesses......Page 320
Problems......Page 323
Chapter 13: Combining Methods: Ensembles and Uplift Modeling......Page 326
Why Ensembles Can Improve Predictive Power......Page 327
Simple Averaging......Page 329
Boosting......Page 330
13.2 Uplift (Persuasion) Modeling......Page 331
Uplift......Page 332
Gathering the Data......Page 333
Modeling Individual Uplift......Page 335
13.3 Summary......Page 337
Problems......Page 338
Part V: Mining Relationships among Records......Page 340
Chapter 14: Association Rules and Collaborative Filtering......Page 342
Example 1: Synthetic Data on Purchases of Phone Faceplates......Page 343
Generating Candidate Rules......Page 345
Selecting Strong Rules......Page 346
Data Format......Page 348
The Process of Rule Selection......Page 349
Interpreting the Results......Page 351
Rules and Chance......Page 352
Example 2: Rules for Similar Book Purchases......Page 354
Data Type and Format......Page 356
Example 3: Netflix Prize Contest......Page 357
User-Based Collaborative Filtering: People Like YouŽ......Page 358
Item-Based Collaborative Filtering......Page 361
Collaborative Filtering vs. Association Rules......Page 362
14.3 Summary......Page 364
Problems......Page 366
Chapter 15: Cluster Analysis......Page 370
15.1 Introduction......Page 371
Example: Public Utilities......Page 372
Euclidean Distance......Page 374
Other Distance Measures for Numerical Data......Page 375
Distance Measures for Categorical Data......Page 377
Distance Measures for Mixed Data......Page 378
Centroid Distance......Page 379
15.4 Hierarchical (Agglomerative) Clustering......Page 381
Complete Linkage......Page 382
Wards Method......Page 383
Dendrograms: Displaying Clustering Process and Results......Page 384
Validating Clusters......Page 386
Limitations of Hierarchical Clustering......Page 387
15.5 Non-hierarchical Clustering: The k-Means Algorithm......Page 388
Initial Partition into k Clusters......Page 390
Problems......Page 394
Part VI: Forecasting Time Series......Page 396
16.1 Introduction......Page 398
Combining Methods......Page 400
Example: Ridership on Amtrak Trains......Page 401
16.5 Data Partitioning and Performance Evaluation......Page 405
Generating Future Forecasts......Page 406
Problems......Page 408
Linear Trend......Page 411
Exponential Trend......Page 413
Polynomial Trend......Page 416
17.2 A Model with Seasonality......Page 417
17.3 A Model with Trend and Seasonality......Page 418
Computing Autocorrelation......Page 421
Improving Forecasts by Integrating Autocorrelation Information......Page 423
Evaluating Predictability......Page 427
Problems......Page 430
18.1 Introduction......Page 441
Centered Moving Average for Visualization......Page 442
Trailing Moving Average for Forecasting......Page 443
Choosing Window Width (w)......Page 445
18.3 Simple Exponential Smoothing......Page 448
Relation between Moving Average and Simple Exponential Smoothing......Page 449
18.4 Advanced Exponential Smoothing......Page 450
Series with a Trend and Seasonality......Page 451
Series with Seasonality (No Trend)......Page 452
Problems......Page 454
Part VII: Data Analytics......Page 462
19.1 Introduction......Page 464
19.2 Directed vs. Undirected Networks......Page 465
19.3 Visualizing and Analyzing Networks......Page 466
Graph Layout......Page 468
Adjacency List......Page 469
Using Network Data in Classification and Prediction......Page 470
19.4 Social Data Metrics and Taxonomy......Page 471
Egocentric Network......Page 472
Network Metrics......Page 473
19.5 Using Network Metrics in Prediction and Classification......Page 474
Link Prediction......Page 475
Entity Resolution......Page 476
Collaborative Filtering......Page 478
19.6 Advantages and Disadvantages......Page 480
Problems......Page 482
20.1 Introduction......Page 484
20.2 The Spreadsheet Representation of Text: Bag-of-WordsŽ......Page 485
20.3 Bag-of-Words vs. Meaning Extraction at Document Level......Page 486
Tokenization......Page 487
Presence/Absence vs. Frequency......Page 488
From Terms to Concepts: Latent Semantic Indexing......Page 489
20.5 Implementing Data Mining Methods......Page 490
Importing and Labeling the Records......Page 491
Tokenization......Page 492
Text Processing and Reduction......Page 493
Producing a Concept Matrix......Page 494
Labeling the Documents......Page 495
20.7 Summary......Page 496
Problems......Page 498
Part VIII: Cases......Page 500
The Book Industry......Page 502
Database Marketing at Charles......Page 503
Data Mining Techniques......Page 506
Assignment......Page 507
Data......Page 511
Assignment......Page 515
Data......Page 516
Assignment......Page 518
Political Targeting......Page 520
Uplift......Page 521
Assignment......Page 522
Assignment......Page 524
Business Situation......Page 525
Measuring Brand Loyalty......Page 526
Appendix......Page 528
Assignment......Page 529
Background......Page 531
Assignment......Page 532
Predicting Corporate Bankruptcy......Page 533
Assignment......Page 534
Assignment Goal......Page 536
Tips and Suggested Steps......Page 537
References......Page 538
Data Files Used in the Book......Page 540
Index......Page 542
End User License Agreement......Page 549