Data Mining for Business Analytics: Concepts, Techniques, and Applications in R

DATA MINING FOR BUSINESS ANALYTICS......Page 3
Contents......Page 9
Foreword by Gareth James......Page 21
Foreword by Ravi Bapna......Page 23
Preface to the R Edition......Page 25
Acknowledgments......Page 29
PART I PRELIMINARIES......Page 33
  1.1 What Is Business Analytics?......Page 35
  1.3 Data Mining and Related Terms......Page 37
  1.4 Big Data......Page 38
  1.5 Data Science......Page 39
  1.6 Why Are There So Many Different Methods?......Page 40
  1.7 Terminology and Notation......Page 41
   Order of Topics......Page 43
  2.1 Introduction......Page 47
   Association Rules and Recommendation Systems......Page 48
   Data Exploration and Visualization......Page 49
   Supervised and Unsupervised Learning......Page 50
  2.3 The Steps in Data Mining......Page 51
   Predicting Home Values in the West Roxbury Neighborhood......Page 53
   Loading and Looking at the Data in R......Page 54
   Sampling from a Database......Page 56
   Oversampling Rare Events in Classification Tasks......Page 57
   Preprocessing and Cleaning the Data......Page 58
   Overfitting......Page 65
   Creation and Use of Data Partitions......Page 67
  2.6 Building a Predictive Model......Page 70
   Modeling Process......Page 71
  2.8 Automating Data Mining Solutions......Page 75
   Data Mining Software: The State of the Market (by Herb Edelstein)......Page 77
  Problems......Page 81
PART II DATA EXPLORATION AND DIMENSION REDUCTION......Page 85
  3.1 Uses of Data Visualization......Page 87
   Example 1: Boston Housing Data......Page 89
  3.3 Basic Charts: Bar Charts, Line Graphs, and Scatter Plots......Page 91
   Distribution Plots: Boxplots and Histograms......Page 93
   Heatmaps: Visualizing Correlations and Missing Values......Page 96
   Adding Variables: Color, Size, Shape, Multiple Panels, and Animation......Page 99
   Manipulations: Rescaling, Aggregation and Hierarchies, Zooming, Filtering......Page 102
   Scaling up to Large Datasets......Page 106
   Multivariate Plot: Parallel Coordinates Plot......Page 107
   Interactive Visualization......Page 109
   Visualizing Networked Data......Page 112
   Visualizing Hierarchical Data: Treemaps......Page 114
   Visualizing Geographical Data: Map Charts......Page 115
   Time Series Forecasting......Page 118
   Unsupervised Learning......Page 119
   Problems......Page 120
  4.1 Introduction......Page 123
  4.3 Practical Considerations......Page 124
   Example 1: House Prices in Boston......Page 125
   Summary Statistics......Page 126
   Aggregation and Pivot Tables......Page 128
  4.5 Correlation Analysis......Page 129
  4.7 Converting a Categorical Variable to a Numerical Variable......Page 131
   Example 2: Breakfast Cereals......Page 133
   Principal Components......Page 138
   Normalizing the Data......Page 139
   Using Principal Components for Classification and Prediction......Page 141
  4.10 Dimension Reduction Using Classification and Regression Trees......Page 143
  Problems......Page 144
PART III PERFORMANCE EVALUATION......Page 147
  5.1 Introduction......Page 149
   Naive Benchmark: The Average......Page 150
   Prediction Accuracy Measures......Page 151
   Lift Chart......Page 153
  5.3 Judging Classifier Performance......Page 154
   The Confusion (Classification) Matrix......Page 156
   Accuracy Measures......Page 158
   Propensities and Cutoff for Classification......Page 159
   Performance in Case of Unequal Importance of Classes......Page 163
   Asymmetric Misclassification Costs......Page 165
   Generalization to More Than Two Classes......Page 167
   Lift Charts for Binary Data......Page 168
   Decile Lift Charts......Page 170
   Lift Charts Incorporating Costs and Benefits......Page 171
  5.5 Oversampling......Page 172
   Evaluating Model Performance if Only Oversampled Validation Set Exists......Page 176
  Problems......Page 179
PART IV PREDICTION AND CLASSIFICATION METHODS......Page 183
  6.1 Introduction......Page 185
  6.2 Explanatory vs. Predictive Modeling......Page 186
   Example: Predicting the Price of Used Toyota Corolla Cars......Page 188
   Reducing the Number of Predictors......Page 193
   How to Reduce the Number of Predictors......Page 194
  Problems......Page 201
   Determining Neighbors......Page 205
   Classification Rule......Page 206
   Example: Riding Mowers......Page 207
   Choosing k......Page 208
   Setting the Cutoff Value......Page 211
  7.2 k-NN for a Numerical Outcome......Page 212
  7.3 Advantages and Shortcomings of k-NN Algorithms......Page 214
  Problems......Page 216
  8.1 Introduction......Page 219
   Example 1: Predicting Fraudulent Financial Reporting......Page 220
  8.2 Applying the Full (Exact) Bayesian Classifier......Page 221
   Practical Difficulty with the Complete (Exact) Bayes Procedure......Page 222
   Solution: Naive Bayes......Page 223
   Using the Cutoff Probability Method......Page 224
   Example 2: Predicting Fraudulent Financial Reports, Two Predictors......Page 225
   Example 3: Predicting Delayed Flights......Page 226
  8.3 Advantages and Shortcomings of the Naive Bayes Classifier......Page 231
  Problems......Page 234
  9.1 Introduction......Page 237
   Example 1: Riding Mowers......Page 239
   Measures of Impurity......Page 242
   Classifying a New Record......Page 246
   Example 2: Acceptance of Personal Loan......Page 247
  9.4 Avoiding Overfitting......Page 248
   Stopping Tree Growth: Conditional Inference Trees......Page 253
   Cross-Validation......Page 254
   Best-Pruned Tree......Page 256
  9.5 Classification Rules from Trees......Page 258
  9.7 Regression Trees......Page 259
   Measuring Impurity......Page 260
   Random Forests......Page 261
   Boosted Trees......Page 263
  9.9 Advantages and Weaknesses of a Tree......Page 264
  Problems......Page 266
  10.1 Introduction......Page 269
  10.2 The Logistic Regression Model......Page 271
  10.3 Example: Acceptance of Personal Loan......Page 272
   Model with a Single Predictor......Page 273
   Estimating the Logistic Model from Data: Computing Parameter Estimates......Page 275
   Interpreting Results in Terms of Odds (for a Profiling Goal)......Page 276
  10.4 Evaluating Classification Performance......Page 279
   Variable Selection......Page 280
  10.5 Example of Complete Analysis: Predicting Delayed Flights......Page 282
   Data Preprocessing......Page 283
   Model Performance......Page 286
   Variable Selection......Page 289
   Appendix A: Why Linear Regression Is Problematic for a Categorical Outcome......Page 291
   Appendix B: Evaluating Explanatory Power......Page 293
   Appendix C: Logistic Regression for More Than Two Classes......Page 296
  Problems......Page 300
  11.1 Introduction......Page 303
  11.2 Concept and Structure of a Neural Network......Page 304
   Example 1: Tiny Dataset......Page 305
   Computing Output of Nodes......Page 306
   Preprocessing the Data......Page 309
   Training the Model......Page 310
   Example 2: Classifying Accident Severity......Page 314
   Using the Output for Prediction and Classification......Page 315
  11.4 Required User Input......Page 317
  11.5 Exploring the Relationship Between Predictors and Outcome......Page 319
  11.6 Advantages and Weaknesses of Neural Networks......Page 320
  Problems......Page 322
  12.1 Introduction......Page 325
   Example 2: Personal Loan Acceptance......Page 326
  12.2 Distance of a Record from a Class......Page 328
  12.3 Fisher’s Linear Classification Functions......Page 329
  12.4 Classification Performance of Discriminant Analysis......Page 332
  12.6 Unequal Misclassification Costs......Page 334
   Example 3: Medical Dispatch to Accident Scenes......Page 335
  12.8 Advantages and Weaknesses......Page 338
  Problems......Page 339
  13.1 Ensembles......Page 343
   Why Ensembles Can Improve Predictive Power......Page 344
   Simple Averaging......Page 346
   Advantages and Weaknesses of Ensembles......Page 347
  13.2 Uplift (Persuasion) Modeling......Page 349
   Uplift......Page 350
   Gathering the Data......Page 351
   A Simple Model......Page 352
   Modeling Individual Uplift......Page 353
   Using the Results of an Uplift Model......Page 354
  13.3 Summary......Page 356
  Problems......Page 357
PART V MINING RELATIONSHIPS AMONG RECORDS......Page 359
  14.1 Association Rules......Page 361
   Generating Candidate Rules......Page 362
   Selecting Strong Rules......Page 365
   Data Format......Page 367
   The Process of Rule Selection......Page 368
   Interpreting the Results......Page 369
   Rules and Chance......Page 371
   Example 2: Rules for Similar Book Purchases......Page 372
  14.2 Collaborative Filtering......Page 374
   Example 3: Netflix Prize Contest......Page 375
   User-Based Collaborative Filtering: “People Like You”......Page 376
   Item-Based Collaborative Filtering......Page 379
   Advantages and Weaknesses of Collaborative Filtering......Page 380
   Collaborative Filtering vs. Association Rules......Page 381
  14.3 Summary......Page 383
  Problems......Page 384
  15.1 Introduction......Page 389
   Example: Public Utilities......Page 391
   Euclidean Distance......Page 393
   Other Distance Measures for Numerical Data......Page 394
   Distance Measures for Categorical Data......Page 397
   Maximum Distance......Page 398
   Centroid Distance......Page 399
  15.4 Hierarchical (Agglomerative) Clustering......Page 400
   Single Linkage......Page 401
   Ward’s Method......Page 402
   Dendrograms: Displaying Clustering Process and Results......Page 403
   Validating Clusters......Page 405
   Limitations of Hierarchical Clustering......Page 407
  15.5 Non-Hierarchical Clustering: The k-Means Algorithm......Page 408
   Choosing the Number of Clusters (k)......Page 409
  Problems......Page 414
PART VI FORECASTING TIME SERIES......Page 417
  16.1 Introduction......Page 419
   Combining Methods......Page 421
   Example: Ridership on Amtrak Trains......Page 422
   Benchmark Performance: Naive Forecasts......Page 427
   Generating Future Forecasts......Page 428
  Problems......Page 430
   Linear Trend......Page 433
   Exponential Trend......Page 437
  17.2 A Model with Seasonality......Page 439
  17.3 A Model with Trend and Seasonality......Page 443
  17.4 Autocorrelation and ARIMA Models......Page 444
   Computing Autocorrelation......Page 445
   Improving Forecasts by Integrating Autocorrelation Information......Page 448
   Evaluating Predictability......Page 452
  Problems......Page 454
  18.1 Introduction......Page 465
   Centered Moving Average for Visualization......Page 466
   Trailing Moving Average for Forecasting......Page 467
  18.3 Simple Exponential Smoothing......Page 471
   Relation Between Moving Average and Simple Exponential Smoothing......Page 472
   Series with a Trend......Page 474
   Series with Seasonality (No Trend)......Page 475
  Problems......Page 478
PART VII DATA ANALYTICS......Page 485
  19.1 Introduction......Page 487
  19.2 Directed vs. Undirected Networks......Page 489
   Graph Layout......Page 490
   Edge List......Page 492
   Using Network Data in Classification and Prediction......Page 493
  19.4 Social Data Metrics and Taxonomy......Page 494
   Egocentric Network......Page 495
   Network Metrics......Page 497
   Entity Resolution......Page 499
   Collaborative Filtering......Page 500
  19.6 Collecting Social Network Data with R......Page 503
  19.7 Advantages and Disadvantages......Page 506
  Problems......Page 508
  20.1 Introduction......Page 511
  20.2 The Tabular Representation of Text: Term-Document Matrix and “Bag-of-Words”......Page 512
  20.3 Bag-of-Words vs. Meaning Extraction at Document Level......Page 513
  20.4 Preprocessing the Text......Page 514
   Tokenization......Page 516
   Text Reduction......Page 517
   Term Frequency–Inverse Document Frequency (TF-IDF)......Page 519
   From Terms to Concepts: Latent Semantic Indexing......Page 520
  20.5 Implementing Data Mining Methods......Page 521
   Importing and Labeling the Records......Page 522
   Producing a Concept Matrix......Page 523
   Prediction......Page 524
  20.7 Summary......Page 526
  Problems......Page 527
PART VIII CASES......Page 529
   The Book Industry......Page 531
   Database Marketing at Charles......Page 532
   Data Mining Techniques......Page 534
   Assignment......Page 536
   Background......Page 537
   Data......Page 538
   Assignment......Page 539
   Data......Page 542
   Assignment......Page 544
   Predictive Analytics Arrives in US Politics......Page 545
   Uplift......Page 546
   Data......Page 547
   Assignment......Page 548
   Assignment......Page 549
   Business Situation......Page 550
   Measuring Brand Loyalty......Page 551
   Background......Page 553
   Data......Page 554
   Assignment......Page 555
   Assignment......Page 556
   Predicting Corporate Bankruptcy......Page 557
   Assignment......Page 558
   Assignment Goal......Page 560
   Tips and Suggested Steps......Page 561
References......Page 563
Data Files Used in the Book......Page 565
Index......Page 567
EULA......Page 577