Textbook introduction to data mining

Contents Preface Introduction 1.1 What Is Data Mining? 1.2 Motivating Challenges 1.3 The Origins of Data Mining 1.4 Data Mining Tasks 1.5 Scope and Organization of the Book 1.6 Bibliographic Notes 1.7 Exercises vii Data 2.1 Types of Data 2.1.1 Attributes and Measurement 2.1.2 Types of Data Sets 2.2 Data Quality 2.2.1 Measurement and Data Collection Issues 2.2.2 Issues Related to Applications 2.3 Data Preprocessing 2.3.1 Aggregation 2.3.2 Sampling 2.3.3 Dimensionality Reduction 2.3.4 Feature Subset Selection 2.3.5 Feature Creation 2.3.6 Discretization and Binarization 2.3.7 Variable Transformation 2.4 Measures of Similarity and Dissimilarity 2.4.1 Basics 2.4.2 Similarity and Dissimilarity between Simple 2.4.3 Dissimilarities between Data Objects 2.4.4 Similarities between Data Objects 11 13 16 Attributes 19 22 23 29 36 37 43 44 45 47 50 52 55 57 63 65 66 67 69 72 xiv Contents 73 80 83 84 88 Exploring Data 3.1 The Iris Data Set 3.2 Summary Statistics 3.2.1 Frequencies and the Mode 3.2.2 Percentiles 3.2.3 Measures of Location: Mean and Median 3.2.4 Measures of Spread: Range and Variance 3.2.5 Multivariate Summary Statistics 3.2.6 Other Ways to Summarize the Data 3.3 Visualization 3.3.1 Motivations for Visualization 3.3.2 General Concepts 3.3.3 Techniques 3.3.4 Visualizing Higher-Dimensional Data 3.3.5 Do’s and Don’ts 3.4 OLAP and Multidimensional Data Analysis 3.4.1 Representing Iris Data as a Multidimensional Array 3.4.2 Multidimensional Data: The General Case 3.4.3 Analyzing Multidimensional Data 3.4.4 Final Comments on Multidimensional Data Analysis 3.5 Bibliographic Notes 3.6 Exercises 97 98 98 99 100 101 102 104 105 105 105 106 110 124 130 131 131 133 135 139 139 141 Classification: Basic Concepts, Decision Trees, and Model Evaluation 4.1 Preliminaries 4.2 General Approach to Solving a Classification Problem 4.3 Decision Tree Induction 4.3.1 How a Decision Tree Works 4.3.2 How to Build a Decision Tree 4.3.3 Methods for Expressing Attribute Test Conditions 4.3.4 Measures for Selecting the Best Split 4.3.5 Algorithm for Decision Tree Induction 4.3.6 An Example: Web Robot Detection 145 146 148 150 150 151 155 158 164 166 2.5 2.6 2.4.5 Examples of Proximity Measures 2.4.6 Issues in Proximity Calculation 2.4.7 Selecting the Right Proximity Measure Bibliographic Notes Exercises Contents 4.4 4.5 4.6 4.7 4.8 4.3.7 Characteristics of Decision Tree Induction Model Overfitting 4.4.1 Overfitting Due to Presence of Noise 4.4.2 Overfitting Due to Lack of Representative Samples 4.4.3 Overfitting and the Multiple Comparison Procedure 4.4.4 Estimation of Generalization Errors 4.4.5 Handling Overfitting in Decision Tree Induction Evaluating the Performance of a Classifier 4.5.1 Holdout Method 4.5.2 Random Subsampling 4.5.3 Cross-Validation 4.5.4 Bootstrap Methods for Comparing Classifiers 4.6.1 Estimating a Confidence Interval for Accuracy 4.6.2 Comparing the Performance of Two Models 4.6.3 Comparing the Performance of Two Classifiers Bibliographic Notes Exercises Classification: Alternative Techniques 5.1 Rule-Based Classifier 5.1.1 How a Rule-Based Classifier Works 5.1.2 Rule-Ordering Schemes 5.1.3 How to Build a Rule-Based Classifier 5.1.4 Direct Methods for Rule Extraction 5.1.5 Indirect Methods for Rule Extraction 5.1.6 Characteristics of Rule-Based Classifiers 5.2 Nearest-Neighbor classifiers 5.2.1 Algorithm 5.2.2 Characteristics of Nearest-Neighbor Classifiers 5.3 Bayesian Classifiers 5.3.1 Bayes Theorem 5.3.2 Using the Bayes Theorem for Classication 5.3.3 Naăve Bayes Classier 5.3.4 Bayes Error Rate 5.3.5 Bayesian Belief Networks 5.4 Artificial Neural Network (ANN) 5.4.1 Perceptron 5.4.2 Multilayer Artificial Neural Network 5.4.3 Characteristics of ANN xv 168 172 175 177 178 179 184 186 186 187 187 188 188 189 191 192 193 198 207 207 209 211 212 213 221 223 223 225 226 227 228 229 231 238 240 246 247 251 255 xvi Contents 5.5 Support Vector Machine (SVM) 5.5.1 Maximum Margin Hyperplanes 5.5.2 Linear SVM: Separable Case 5.5.3 Linear SVM: Nonseparable Case 5.5.4 Nonlinear SVM 5.5.5 Characteristics of SVM 5.6 Ensemble Methods 5.6.1 Rationale for Ensemble Method 5.6.2 Methods for Constructing an Ensemble Classifier 5.6.3 Bias-Variance Decomposition 5.6.4 Bagging 5.6.5 Boosting 5.6.6 Random Forests 5.6.7 Empirical Comparison among Ensemble Methods 5.7 Class Imbalance Problem 5.7.1 Alternative Metrics 5.7.2 The Receiver Operating Characteristic Curve 5.7.3 Cost-Sensitive Learning 5.7.4 Sampling-Based Approaches 5.8 Multiclass Problem 5.9 Bibliographic Notes 5.10 Exercises 256 256 259 266 270 276 276 277 278 281 283 285 290 294 294 295 298 302 305 306 309 315 Association Analysis: Basic Concepts and Algorithms 6.1 Problem Definition 6.2 Frequent Itemset Generation 6.2.1 The Apriori Principle 6.2.2 Frequent Itemset Generation in the Apriori Algorithm 6.2.3 Candidate Generation and Pruning 6.2.4 Support Counting 6.2.5 Computational Complexity 6.3 Rule Generation 6.3.1 Confidence-Based Pruning 6.3.2 Rule Generation in Apriori Algorithm 6.3.3 An Example: Congressional Voting Records 6.4 Compact Representation of Frequent Itemsets 6.4.1 Maximal Frequent Itemsets 6.4.2 Closed Frequent Itemsets 6.5 Alternative Methods for Generating Frequent Itemsets 6.6 FP-Growth Algorithm 327 328 332 333 335 338 342 345 349 350 350 352 353 354 355 359 363 Contents 6.6.1 FP-Tree Representation 6.6.2 Frequent Itemset Generation in FP-Growth Algorithm 6.7 Evaluation of Association Patterns 6.7.1 Objective Measures of Interestingness 6.7.2 Measures beyond Pairs of Binary Variables 6.7.3 Simpson’s Paradox 6.8 Effect of Skewed Support Distribution 6.9 Bibliographic Notes 6.10 Exercises xvii 363 366 370 371 382 384 386 390 404 Association Analysis: Advanced Concepts 7.1 Handling Categorical Attributes 7.2 Handling Continuous Attributes 7.2.1 Discretization-Based Methods 7.2.2 Statistics-Based Methods 7.2.3 Non-discretization Methods 7.3 Handling a Concept Hierarchy 7.4 Sequential Patterns 7.4.1 Problem Formulation 7.4.2 Sequential Pattern Discovery 7.4.3 Timing Constraints 7.4.4 Alternative Counting Schemes 7.5 Subgraph Patterns 7.5.1 Graphs and Subgraphs 7.5.2 Frequent Subgraph Mining 7.5.3 Apriori -like Method 7.5.4 Candidate Generation 7.5.5 Candidate Pruning 7.5.6 Support Counting 7.6 Infrequent Patterns 7.6.1 Negative Patterns 7.6.2 Negatively Correlated Patterns 7.6.3 Comparisons among Infrequent Patterns, Negative Patterns, and Negatively Correlated Patterns 7.6.4 Techniques for Mining Interesting Infrequent Patterns 7.6.5 Techniques Based on Mining Negative Patterns 7.6.6 Techniques Based on Support Expectation 7.7 Bibliographic Notes 7.8 Exercises 415 415 418 418 422 424 426 429 429 431 436 439 442 443 444 447 448 453 457 457 458 458 460 461 463 465 469 473 xviii Contents Cluster Analysis: Basic Concepts and Algorithms 8.1 Overview 8.1.1 What Is Cluster Analysis? 8.1.2 Different Types of Clusterings 8.1.3 Different Types of Clusters 8.2 K-means 8.2.1 The Basic K-means Algorithm 8.2.2 K-means: Additional Issues 8.2.3 Bisecting K-means 8.2.4 K-means and Different Types of Clusters 8.2.5 Strengths and Weaknesses 8.2.6 K-means as an Optimization Problem 8.3 Agglomerative Hierarchical Clustering 8.3.1 Basic Agglomerative Hierarchical Clustering Algorithm 8.3.2 Specific Techniques 8.3.3 The Lance-Williams Formula for Cluster Proximity 8.3.4 Key Issues in Hierarchical Clustering 8.3.5 Strengths and Weaknesses 8.4 DBSCAN 8.4.1 Traditional Density: Center-Based Approach 8.4.2 The DBSCAN Algorithm 8.4.3 Strengths and Weaknesses 8.5 Cluster Evaluation 8.5.1 Overview 8.5.2 Unsupervised Cluster Evaluation Using Cohesion and Separation 8.5.3 Unsupervised Cluster Evaluation Using the Proximity Matrix 8.5.4 Unsupervised Evaluation of Hierarchical Clustering 8.5.5 Determining the Correct Number of Clusters 8.5.6 Clustering Tendency 8.5.7 Supervised Measures of Cluster Validity 8.5.8 Assessing the Significance of Cluster Validity Measures 8.6 Bibliographic Notes 8.7 Exercises 487 490 490 491 493 496 497 506 508 510 510 513 515 516 518 524 524 526 526 527 528 530 532 533 536 542 544 546 547 548 553 555 559 Cluster Analysis: Additional Issues and Algorithms 569 9.1 Characteristics of Data, Clusters, and Clustering Algorithms 570 9.1.1 Example: Comparing K-means and DBSCAN 570 9.1.2 Data Characteristics 571 Contents xix 9.1.3 Cluster Characteristics 9.1.4 General Characteristics of Clustering Algorithms Prototype-Based Clustering 9.2.1 Fuzzy Clustering 9.2.2 Clustering Using Mixture Models 9.2.3 Self-Organizing Maps (SOM) Density-Based Clustering 9.3.1 Grid-Based Clustering 9.3.2 Subspace Clustering 9.3.3 DENCLUE: A Kernel-Based Scheme for Density-Based Clustering Graph-Based Clustering 9.4.1 Sparsification 9.4.2 Minimum Spanning Tree (MST) Clustering 9.4.3 OPOSSUM: Optimal Partitioning of Sparse Similarities Using METIS 9.4.4 Chameleon: Hierarchical Clustering with Dynamic Modeling 9.4.5 Shared Nearest Neighbor Similarity 9.4.6 The Jarvis-Patrick Clustering Algorithm 9.4.7 SNN Density 9.4.8 SNN Density-Based Clustering Scalable Clustering Algorithms 9.5.1 Scalability: General Issues and Approaches 9.5.2 BIRCH 9.5.3 CURE Which Clustering Algorithm? Bibliographic Notes Exercises 573 575 577 577 583 594 600 601 604 10 Anomaly Detection 10.1 Preliminaries 10.1.1 Causes of Anomalies 10.1.2 Approaches to Anomaly Detection 10.1.3 The Use of Class Labels 10.1.4 Issues 10.2 Statistical Approaches 10.2.1 Detecting Outliers in a Univariate Normal Distribution 10.2.2 Outliers in a Multivariate Normal Distribution 10.2.3 A Mixture Model Approach for Anomaly Detection 651 653 653 654 655 656 658 659 661 662 9.2 9.3 9.4 9.5 9.6 9.7 9.8 608 612 613 614 616 616 622 625 627 629 630 630 633 635 639 643 647 xx Contents 10.2.4 Strengths and Weaknesses 10.3 Proximity-Based Outlier Detection 10.3.1 Strengths and Weaknesses 10.4 Density-Based Outlier Detection 10.4.1 Detection of Outliers Using Relative Density 10.4.2 Strengths and Weaknesses 10.5 Clustering-Based Techniques 10.5.1 Assessing the Extent to Which an Object Belongs to a Cluster 10.5.2 Impact of Outliers on the Initial Clustering 10.5.3 The Number of Clusters to Use 10.5.4 Strengths and Weaknesses 10.6 Bibliographic Notes 10.7 Exercises 665 666 666 668 669 670 671 672 674 674 674 675 680 Appendix A Linear Algebra A.1 Vectors A.1.1 Definition A.1.2 Vector Addition and Multiplication by a Scalar A.1.3 Vector Spaces A.1.4 The Dot Product, Orthogonality, and Orthogonal Projections A.1.5 Vectors and Data Analysis A.2 Matrices A.2.1 Matrices: Definitions A.2.2 Matrices: Addition and Multiplication by a Scalar A.2.3 Matrices: Multiplication A.2.4 Linear Transformations and Inverse Matrices A.2.5 Eigenvalue and Singular Value Decomposition A.2.6 Matrices and Data Analysis A.3 Bibliographic Notes 685 685 685 685 687 688 690 691 691 692 693 695 697 699 700 Appendix B Dimensionality Reduction B.1 PCA and SVD B.1.1 Principal Components Analysis (PCA) B.1.2 SVD B.2 Other Dimensionality Reduction Techniques B.2.1 Factor Analysis B.2.2 Locally Linear Embedding (LLE) B.2.3 Multidimensional Scaling, FastMap, and 701 701 701 706 708 708 710 712 ISOMAP ...Contents Preface Introduction 1.1 What Is Data Mining? 1.2 Motivating Challenges 1.3 The Origins of Data Mining 1.4 Data Mining Tasks 1.5 Scope... Data 2.1 Types of Data 2.1.1 Attributes and Measurement 2.1.2 Types of Data Sets 2.2 Data Quality 2.2.1 Measurement and Data. .. Higher-Dimensional Data 3.3.5 Do’s and Don’ts 3.4 OLAP and Multidimensional Data Analysis 3.4.1 Representing Iris Data as a Multidimensional Array 3.4.2 Multidimensional Data: