|
% Usage: |
|
% * download https://web.stanford.edu/~hastie/ElemStatLearn/printings/ESLII_print12.pdf |
|
% as 'in.pdf' |
|
% * store this file as 'out.tex', and compile as 'pdflatex out.tex' |
|
% * rename output file to e.g. |
|
% 'Hastie, Tibshirani and Friedman - The Elements of Statistical Learning.pdf' |
|
|
|
|
|
\documentclass{article} |
|
\usepackage[utf8]{inputenc} |
|
|
|
\usepackage{geometry} |
|
%\geometry{papersize={170mm,257mm}} |
|
% You may uncomment the above line to create the book in the original size. |
|
% Otherwise, the output page size will be the defaul letter or A4, which |
|
% I prefer (extra margins for notes) |
|
|
|
\usepackage{pdfpages} |
|
\usepackage[ |
|
pdfpagelabels=true, |
|
pdftitle={The Elements of Statistical Learning (2nd edition)}, |
|
pdfauthor={Trevor Hastie, Robert Tibshirani, Jerome Friedman}, |
|
pdfsubject={Mathematical statistics, Machine Learning, Data Mining}, |
|
pdfkeywords={supervised learning, machine learning, linear methods, prediction}, |
|
unicode=true, |
|
]{hyperref} |
|
\usepackage{bookmark} |
|
|
|
\begin{document} |
|
|
|
\pagenumbering{roman} |
|
\setcounter{page}{3} |
|
\includepdf[pages={1, {}}]{in.pdf} |
|
\includepdf[pages=2-19]{in.pdf} |
|
|
|
\pagenumbering{arabic} |
|
\setcounter{page}{1} |
|
\includepdf[pages=20-]{in.pdf} |
|
|
|
\bookmark[page=1,level=0]{Cover} |
|
\bookmark[page=5,level=0]{Preface to the Second Edition} |
|
\bookmark[page=9,level=0]{Preface to the First Edition} |
|
\bookmark[page=11,level=0]{Contents} |
|
|
|
\bookmark[page=21,level=0]{1 Introduction} |
|
\bookmark[page=29,level=0]{2 Overview of Supervised Learning} |
|
\bookmark[page=29,level=1]{2.1 Introduction} |
|
\bookmark[page=29,level=1]{2.2 Variable Types and Terminology} |
|
\bookmark[page=31,level=1]{2.3 Two Simple Approaches to Prediction: Least Squares and Nearest Neighbors} |
|
\bookmark[page=31,level=2]{2.3.1 Linear Models and Least Squares} |
|
\bookmark[page=34,level=2]{2.3.2 Nearest-Neighbor Methods} |
|
\bookmark[page=36,level=2]{2.3.3 From Least Squares to Nearest Neighbors} |
|
\bookmark[page=38,level=1]{2.4 Statistical Decision Theory} |
|
\bookmark[page=42,level=1]{2.5 Local Methods in High Dimensions} |
|
\bookmark[page=48,level=1]{2.6 Statistical Models, Supervised Learning and Function Approximation} |
|
\bookmark[page=48,level=2]{2.6.1 A Statistical Model for the Joint Distribution Pr(X,Y)} |
|
\bookmark[page=49,level=2]{2.6.2 Supervised Learning} |
|
\bookmark[page=49,level=2]{2.6.3 Function Approximation} |
|
\bookmark[page=52,level=1]{2.7 Structured Regression Models} |
|
\bookmark[page=52,level=2]{2.7.1 Difficulty of the Problem} |
|
\bookmark[page=53,level=1]{2.8 Classes of Restricted Estimators} |
|
\bookmark[page=54,level=2]{2.8.1 Roughness Penalty and Bayesian Methods} |
|
\bookmark[page=54,level=2]{2.8.2 Kernel Methods and Local Regression} |
|
\bookmark[page=55,level=2]{2.8.3 Basis Functions and Dictionary Methods} |
|
\bookmark[page=57,level=1]{2.9 Model Selection and the Bias–Variance Tradeoff} |
|
\bookmark[page=59,level=1]{Bibliographic Notes} |
|
\bookmark[page=59,level=1]{Exercises} |
|
\bookmark[page=63,level=0]{3 Linear Methods for Regression} |
|
\bookmark[page=63,level=1]{3.1 Introduction} |
|
\bookmark[page=64,level=1]{3.2 Linear Regression Models and Least Squares} |
|
\bookmark[page=69,level=2]{3.2.1 Example: Prostate Cancer} |
|
\bookmark[page=71,level=2]{3.2.2 The Gauss–Markov Theorem} |
|
\bookmark[page=72,level=2]{3.2.3 Multiple Regression from Simple Univariate Regression} |
|
\bookmark[page=76,level=2]{3.2.4 Multiple Outputs} |
|
\bookmark[page=77,level=1]{3.3 Subset Selection} |
|
\bookmark[page=77,level=2]{3.3.1 Best-Subset Selection} |
|
\bookmark[page=78,level=2]{3.3.2 Forward- and Backward-Stepwise Selection} |
|
\bookmark[page=80,level=2]{3.3.3 Forward-Stagewise Regression} |
|
\bookmark[page=81,level=2]{3.3.4 Prostate Cancer Data Example (Continued)} |
|
\bookmark[page=81,level=1]{3.4 Shrinkage Methods} |
|
\bookmark[page=81,level=2]{3.4.1 Ridge Regression} |
|
\bookmark[page=88,level=2]{3.4.2 The Lasso} |
|
\bookmark[page=89,level=2]{3.4.3 Discussion: Subset Selection, Ridge Regression and the Lasso} |
|
\bookmark[page=93,level=2]{3.4.4 Least Angle Regression} |
|
\bookmark[page=99,level=1]{3.5 Methods Using Derived Input Directions} |
|
\bookmark[page=99,level=2]{3.5.1 Principal Components Regression} |
|
\bookmark[page=100,level=2]{3.5.2 Partial Least Squares} |
|
\bookmark[page=102,level=1]{3.6 Discussion: A Comparison of the Selection and Shrinkage Methods} |
|
\bookmark[page=104,level=1]{3.7 Multiple Outcome Shrinkage and Selection} |
|
\bookmark[page=106,level=1]{3.8 More on the Lasso and Related Path Algorithms} |
|
\bookmark[page=106,level=2]{3.8.1 Incremental Forward Stagewise Regression} |
|
\bookmark[page=109,level=2]{3.8.2 Piecewise-Linear Path Algorithms} |
|
\bookmark[page=109,level=2]{3.8.3 The Dantzig Selector} |
|
\bookmark[page=110,level=2]{3.8.4 The Grouped Lasso} |
|
\bookmark[page=111,level=2]{3.8.5 Further Properties of the Lasso} |
|
\bookmark[page=112,level=2]{3.8.6 Pathwise Coordinate Optimization} |
|
\bookmark[page=113,level=1]{3.9 Computational Considerations} |
|
\bookmark[page=114,level=1]{Bibliographic Notes} |
|
\bookmark[page=114,level=1]{Exercises} |
|
\bookmark[page=121,level=0]{4 Linear Methods for Classification} |
|
\bookmark[page=121,level=1]{4.1 Introduction} |
|
\bookmark[page=123,level=1]{4.2 Linear Regression of an Indicator Matrix} |
|
\bookmark[page=126,level=1]{4.3 Linear Discriminant Analysis} |
|
\bookmark[page=132,level=2]{4.3.1 Regularized Discriminant Analysis} |
|
\bookmark[page=133,level=2]{4.3.2 Computations for LDA} |
|
\bookmark[page=133,level=2]{4.3.3 Reduced-Rank Linear Discriminant Analysis} |
|
\bookmark[page=139,level=1]{4.4 Logistic Regression} |
|
\bookmark[page=140,level=2]{4.4.1 Fitting Logistic Regression Models} |
|
\bookmark[page=142,level=2]{4.4.2 Example: South African Heart Disease} |
|
\bookmark[page=144,level=2]{4.4.3 Quadratic Approximations and Inference} |
|
\bookmark[page=145,level=2]{4.4.4 L1 Regularized Logistic Regression} |
|
\bookmark[page=147,level=2]{4.4.5 Logistic Regression or LDA?} |
|
\bookmark[page=149,level=1]{4.5 Separating Hyperplanes} |
|
\bookmark[page=150,level=2]{4.5.1 Rosenblatt’s Perceptron Learning Algorithm} |
|
\bookmark[page=152,level=2]{4.5.2 Optimal Separating Hyperplanes} |
|
\bookmark[page=155,level=1]{Bibliographic Notes} |
|
\bookmark[page=155,level=1]{Exercises} |
|
\bookmark[page=159,level=0]{5 Basis Expansions and Regularization} |
|
\bookmark[page=159,level=1]{5.1 Introduction} |
|
\bookmark[page=161,level=1]{5.2 Piecewise Polynomials and Splines} |
|
\bookmark[page=164,level=2]{5.2.1 Natural Cubic Splines} |
|
\bookmark[page=166,level=2]{5.2.2 Example: South African Heart Disease (Continued)} |
|
\bookmark[page=168,level=2]{5.2.3 Example: Phoneme Recognition} |
|
\bookmark[page=170,level=1]{5.3 Filtering and Feature Extraction} |
|
\bookmark[page=171,level=1]{5.4 Smoothing Splines} |
|
\bookmark[page=173,level=2]{5.4.1 Degrees of Freedom and Smoother Matrices} |
|
\bookmark[page=176,level=1]{5.5 Automatic Selection of the Smoothing Parameters} |
|
\bookmark[page=178,level=2]{5.5.1 Fixing the Degrees of Freedom} |
|
\bookmark[page=178,level=2]{5.5.2 The Bias–Variance Tradeoff} |
|
\bookmark[page=181,level=1]{5.6 Nonparametric Logistic Regression} |
|
\bookmark[page=182,level=1]{5.7 Multidimensional Splines} |
|
\bookmark[page=187,level=1]{5.8 Regularization and Reproducing Kernel Hilbert Spaces} |
|
\bookmark[page=188,level=2]{5.8.1 Spaces of Functions Generated by Kernels} |
|
\bookmark[page=190,level=2]{5.8.2 Examples of RKHS} |
|
\bookmark[page=194,level=1]{5.9 Wavelet Smoothing} |
|
\bookmark[page=196,level=2]{5.9.1 Wavelet Bases and the Wavelet Transform} |
|
\bookmark[page=199,level=2]{5.9.2 Adaptive Wavelet Filtering} |
|
\bookmark[page=201,level=1]{Bibliographic Notes} |
|
\bookmark[page=201,level=1]{Exercises} |
|
\bookmark[page=206,level=1]{Appendix: Computational Considerations for Splines} |
|
\bookmark[page=206,level=1]{Appendix: B-splines} |
|
\bookmark[page=209,level=1]{Appendix: Computations for Smoothing Splines} |
|
\bookmark[page=211,level=0]{6 Kernel Smoothing Methods} |
|
\bookmark[page=212,level=1]{6.1 One-Dimensional Kernel Smoothers} |
|
\bookmark[page=214,level=2]{6.1.1 Local Linear Regression} |
|
\bookmark[page=217,level=2]{6.1.2 Local Polynomial Regression} |
|
\bookmark[page=218,level=1]{6.2 Selecting the Width of the Kernel} |
|
\bookmark[page=220,level=1]{6.3 Local Regression in IRp} |
|
\bookmark[page=221,level=1]{6.4 Structured Local Regression Models in IRp} |
|
\bookmark[page=223,level=2]{6.4.1 Structured Kernels} |
|
\bookmark[page=223,level=2]{6.4.2 Structured Regression Functions} |
|
\bookmark[page=225,level=1]{6.5 Local Likelihood and Other Models} |
|
\bookmark[page=228,level=1]{6.6 Kernel Density Estimation and Classification} |
|
\bookmark[page=228,level=2]{6.6.1 Kernel Density Estimation} |
|
\bookmark[page=230,level=2]{6.6.2 Kernel Density Classification} |
|
\bookmark[page=230,level=2]{6.6.3 The Naive Bayes Classifier} |
|
\bookmark[page=232,level=1]{6.7 Radial Basis Functions and Kernels} |
|
\bookmark[page=234,level=1]{6.8 Mixture Models for Density Estimation and Classification} |
|
\bookmark[page=236,level=1]{6.9 Computational Considerations} |
|
\bookmark[page=236,level=1]{Bibliographic Notes} |
|
\bookmark[page=236,level=1]{Exercises} |
|
\bookmark[page=239,level=0]{7 Model Assessment and Selection} |
|
\bookmark[page=239,level=1]{7.1 Introduction} |
|
\bookmark[page=239,level=1]{7.2 Bias, Variance and Model Complexity} |
|
\bookmark[page=243,level=1]{7.3 The Bias–Variance Decomposition} |
|
\bookmark[page=246,level=2]{7.3.1 Example: Bias–Variance Tradeoff} |
|
\bookmark[page=248,level=1]{7.4 Optimism of the Training Error Rate} |
|
\bookmark[page=250,level=1]{7.5 Estimates of In-Sample Prediction Error} |
|
\bookmark[page=252,level=1]{7.6 The Effective Number of Parameters} |
|
\bookmark[page=253,level=1]{7.7 The Bayesian Approach and BIC} |
|
\bookmark[page=255,level=1]{7.8 Minimum Description Length} |
|
\bookmark[page=257,level=1]{7.9 Vapnik–Chervonenkis Dimension} |
|
\bookmark[page=259,level=2]{7.9.1 Example (Continued)} |
|
\bookmark[page=261,level=1]{7.10 Cross-Validation} |
|
\bookmark[page=261,level=2]{7.10.1 K-Fold Cross-Validation} |
|
\bookmark[page=265,level=2]{7.10.2 The Wrong and Right Way to Do Cross-validation} |
|
\bookmark[page=267,level=2]{7.10.3 Does Cross-Validation Really Work?} |
|
\bookmark[page=269,level=1]{7.11 Bootstrap Methods} |
|
\bookmark[page=272,level=2]{7.11.1 Example (Continued)} |
|
\bookmark[page=274,level=1]{7.12 Conditional or Expected Test Error?} |
|
\bookmark[page=277,level=1]{Bibliographic Notes} |
|
\bookmark[page=277,level=1]{Exercises} |
|
\bookmark[page=281,level=0]{8 Model Inference and Averaging} |
|
\bookmark[page=281,level=1]{8.1 Introduction} |
|
\bookmark[page=281,level=1]{8.2 The Bootstrap and Maximum Likelihood Methods} |
|
\bookmark[page=281,level=2]{8.2.1 A Smoothing Example} |
|
\bookmark[page=285,level=2]{8.2.2 Maximum Likelihood Inference} |
|
\bookmark[page=287,level=2]{8.2.3 Bootstrap versus Maximum Likelihood} |
|
\bookmark[page=287,level=1]{8.3 Bayesian Methods} |
|
\bookmark[page=291,level=1]{8.4 Relationship Between the Bootstrap and Bayesian Inference} |
|
\bookmark[page=292,level=1]{8.5 The EM Algorithm} |
|
\bookmark[page=292,level=2]{8.5.1 Two-Component Mixture Model} |
|
\bookmark[page=296,level=2]{8.5.2 The EM Algorithm in General} |
|
\bookmark[page=297,level=2]{8.5.3 EM as a Maximization–Maximization Procedure} |
|
\bookmark[page=299,level=1]{8.6 MCMC for Sampling from the Posterior} |
|
\bookmark[page=302,level=1]{8.7 Bagging} |
|
\bookmark[page=303,level=2]{8.7.1 Example: Trees with Simulated Data} |
|
\bookmark[page=308,level=1]{8.8 Model Averaging and Stacking} |
|
\bookmark[page=310,level=1]{8.9 Stochastic Search: Bumping} |
|
\bookmark[page=312,level=1]{Bibliographic Notes} |
|
\bookmark[page=313,level=1]{Exercises} |
|
\bookmark[page=315,level=0]{9 Additive Models, Trees, and Related Methods} |
|
\bookmark[page=315,level=1]{9.1 Generalized Additive Models} |
|
\bookmark[page=317,level=2]{9.1.1 Fitting Additive Models} |
|
\bookmark[page=319,level=2]{9.1.2 Example: Additive Logistic Regression} |
|
\bookmark[page=324,level=2]{9.1.3 Summary} |
|
\bookmark[page=325,level=1]{9.2 Tree-Based Methods} |
|
\bookmark[page=325,level=2]{9.2.1 Background} |
|
\bookmark[page=327,level=2]{9.2.2 Regression Trees} |
|
\bookmark[page=328,level=2]{9.2.3 Classification Trees} |
|
\bookmark[page=330,level=2]{9.2.4 Other Issues} |
|
\bookmark[page=333,level=2]{9.2.5 Spam Example (Continued)} |
|
\bookmark[page=337,level=1]{9.3 PRIM: Bump Hunting} |
|
\bookmark[page=340,level=2]{9.3.1 Spam Example (Continued)} |
|
\bookmark[page=341,level=1]{9.4 MARS: Multivariate Adaptive Regression Splines} |
|
\bookmark[page=346,level=2]{9.4.1 Spam Example (Continued)} |
|
\bookmark[page=347,level=2]{9.4.2 Example (Simulated Data)} |
|
\bookmark[page=348,level=2]{9.4.3 Other Issues} |
|
\bookmark[page=349,level=1]{9.5 Hierarchical Mixtures of Experts} |
|
\bookmark[page=352,level=1]{9.6 Missing Data} |
|
\bookmark[page=354,level=1]{9.7 Computational Considerations} |
|
\bookmark[page=354,level=1]{Bibliographic Notes} |
|
\bookmark[page=355,level=1]{Exercises} |
|
\bookmark[page=357,level=0]{10 Boosting and Additive Trees} |
|
\bookmark[page=357,level=1]{10.1 Boosting Methods} |
|
\bookmark[page=360,level=2]{10.1.1 Outline of This Chapter} |
|
\bookmark[page=361,level=1]{10.2 Boosting Fits an Additive Model} |
|
\bookmark[page=362,level=1]{10.3 Forward Stagewise Additive Modeling} |
|
\bookmark[page=363,level=1]{10.4 Exponential Loss and AdaBoost} |
|
\bookmark[page=365,level=1]{10.5 Why Exponential Loss?} |
|
\bookmark[page=366,level=1]{10.6 Loss Functions and Robustness} |
|
\bookmark[page=370,level=1]{10.7 “Off-the-Shelf” Procedures for Data Mining} |
|
\bookmark[page=372,level=1]{10.8 Example: Spam Data} |
|
\bookmark[page=373,level=1]{10.9 Boosting Trees} |
|
\bookmark[page=378,level=1]{10.10 Numerical Optimization via Gradient Boosting} |
|
\bookmark[page=378,level=2]{10.10.1 Steepest Descent} |
|
\bookmark[page=379,level=2]{10.10.2 Gradient Boosting} |
|
\bookmark[page=380,level=2]{10.10.3 Implementations of Gradient Boosting} |
|
\bookmark[page=381,level=1]{10.11 Right-Sized Trees for Boosting} |
|
\bookmark[page=384,level=1]{10.12 Regularization} |
|
\bookmark[page=384,level=2]{10.12.1 Shrinkage} |
|
\bookmark[page=385,level=2]{10.12.2 Subsampling} |
|
\bookmark[page=387,level=1]{10.13 Interpretation} |
|
\bookmark[page=387,level=2]{10.13.1 Relative Importance of Predictor Variables} |
|
\bookmark[page=389,level=2]{10.13.2 Partial Dependence Plots} |
|
\bookmark[page=391,level=1]{10.14 Illustrations} |
|
\bookmark[page=391,level=2]{10.14.1 California Housing} |
|
\bookmark[page=395,level=2]{10.14.2 New Zealand Fish} |
|
\bookmark[page=399,level=2]{10.14.3 Demographics Data} |
|
\bookmark[page=400,level=1]{Bibliographic Notes} |
|
\bookmark[page=404,level=1]{Exercises} |
|
\bookmark[page=409,level=0]{11 Neural Networks} |
|
\bookmark[page=409,level=1]{11.1 Introduction} |
|
\bookmark[page=409,level=1]{11.2 Projection Pursuit Regression} |
|
\bookmark[page=412,level=1]{11.3 Neural Networks} |
|
\bookmark[page=415,level=1]{11.4 Fitting Neural Networks} |
|
\bookmark[page=417,level=1]{11.5 Some Issues in Training Neural Networks} |
|
\bookmark[page=417,level=2]{11.5.1 Starting Values} |
|
\bookmark[page=418,level=2]{11.5.2 Overfitting} |
|
\bookmark[page=418,level=2]{11.5.3 Scaling of the Inputs} |
|
\bookmark[page=420,level=2]{11.5.4 Number of Hidden Units and Layers} |
|
\bookmark[page=420,level=2]{11.5.5 Multiple Minima} |
|
\bookmark[page=421,level=1]{11.6 Example: Simulated Data} |
|
\bookmark[page=424,level=1]{11.7 Example: ZIP Code Data} |
|
\bookmark[page=428,level=1]{11.8 Discussion} |
|
\bookmark[page=429,level=1]{11.9 Bayesian Neural Nets and the NIPS 2003 Challenge} |
|
\bookmark[page=430,level=2]{11.9.1 Bayes, Boosting and Bagging} |
|
\bookmark[page=432,level=2]{11.9.2 Performance Comparisons} |
|
\bookmark[page=434,level=1]{11.10 Computational Considerations} |
|
\bookmark[page=435,level=1]{Bibliographic Notes} |
|
\bookmark[page=435,level=1]{Exercises} |
|
\bookmark[page=437,level=0]{12 Support Vector Machines and Flexible Discriminants} |
|
\bookmark[page=437,level=1]{12.1 Introduction} |
|
\bookmark[page=437,level=1]{12.2 The Support Vector Classifier} |
|
\bookmark[page=440,level=2]{12.2.1 Computing the Support Vector Classifier} |
|
\bookmark[page=441,level=2]{12.2.2 Mixture Example (Continued)} |
|
\bookmark[page=443,level=1]{12.3 Support Vector Machines and Kernels} |
|
\bookmark[page=443,level=2]{12.3.1 Computing the SVM for Classification} |
|
\bookmark[page=446,level=2]{12.3.2 The SVM as a Penalization Method} |
|
\bookmark[page=448,level=2]{12.3.3 Function Estimation and Reproducing Kernels} |
|
\bookmark[page=451,level=2]{12.3.4 SVMs and the Curse of Dimensionality} |
|
\bookmark[page=452,level=2]{12.3.5 A Path Algorithm for the SVM Classifier} |
|
\bookmark[page=454,level=2]{12.3.6 Support Vector Machines for Regression} |
|
\bookmark[page=456,level=2]{12.3.7 Regression and Kernels} |
|
\bookmark[page=458,level=2]{12.3.8 Discussion} |
|
\bookmark[page=458,level=1]{12.4 Generalizing Linear Discriminant Analysis} |
|
\bookmark[page=460,level=1]{12.5 Flexible Discriminant Analysis} |
|
\bookmark[page=464,level=2]{12.5.1 Computing the FDA Estimates} |
|
\bookmark[page=466,level=1]{12.6 Penalized Discriminant Analysis} |
|
\bookmark[page=469,level=1]{12.7 Mixture Discriminant Analysis} |
|
\bookmark[page=471,level=2]{12.7.1 Example: Waveform Data} |
|
\bookmark[page=475,level=1]{Bibliographic Notes} |
|
\bookmark[page=475,level=1]{Exercises} |
|
\bookmark[page=479,level=0]{13 Prototype Methods and Nearest-Neighbors} |
|
\bookmark[page=479,level=1]{13.1 Introduction} |
|
\bookmark[page=479,level=1]{13.2 Prototype Methods} |
|
\bookmark[page=480,level=2]{13.2.1 K-means Clustering} |
|
\bookmark[page=482,level=2]{13.2.2 Learning Vector Quantization} |
|
\bookmark[page=483,level=2]{13.2.3 Gaussian Mixtures} |
|
\bookmark[page=483,level=1]{13.3 k-Nearest-Neighbor Classifiers} |
|
\bookmark[page=488,level=2]{13.3.1 Example: A Comparative Study} |
|
\bookmark[page=490,level=2]{13.3.2 Example: k-Nearest-Neighbors and Image Scene Classification} |
|
\bookmark[page=491,level=2]{13.3.3 Invariant Metrics and Tangent Distance} |
|
\bookmark[page=495,level=1]{13.4 Adaptive Nearest-Neighbor Methods} |
|
\bookmark[page=498,level=2]{13.4.1 Example} |
|
\bookmark[page=499,level=2]{13.4.2 Global Dimension Reduction for Nearest-Neighbors} |
|
\bookmark[page=500,level=1]{13.5 Computational Considerations} |
|
\bookmark[page=501,level=1]{Bibliographic Notes} |
|
\bookmark[page=501,level=1]{Exercises} |
|
\bookmark[page=505,level=0]{14 Unsupervised Learning} |
|
\bookmark[page=505,level=1]{14.1 Introduction} |
|
\bookmark[page=507,level=1]{14.2 Association Rules} |
|
\bookmark[page=508,level=2]{14.2.1 Market Basket Analysis} |
|
\bookmark[page=509,level=2]{14.2.2 The Apriori Algorithm} |
|
\bookmark[page=512,level=2]{14.2.3 Example: Market Basket Analysis} |
|
\bookmark[page=515,level=2]{14.2.4 Unsupervised as Supervised Learning} |
|
\bookmark[page=517,level=2]{14.2.5 Generalized Association Rules} |
|
\bookmark[page=519,level=2]{14.2.6 Choice of Supervised Learning Method} |
|
\bookmark[page=519,level=2]{14.2.7 Example: Market Basket Analysis (Continued)} |
|
\bookmark[page=521,level=1]{14.3 Cluster Analysis} |
|
\bookmark[page=523,level=2]{14.3.1 Proximity Matrices} |
|
\bookmark[page=523,level=2]{14.3.2 Dissimilarities Based on Attributes} |
|
\bookmark[page=525,level=2]{14.3.3 Object Dissimilarity} |
|
\bookmark[page=527,level=2]{14.3.4 Clustering Algorithms} |
|
\bookmark[page=527,level=2]{14.3.5 Combinatorial Algorithms} |
|
\bookmark[page=529,level=2]{14.3.6 K-means} |
|
\bookmark[page=530,level=2]{14.3.7 Gaussian Mixtures as Soft K-means Clustering} |
|
\bookmark[page=532,level=2]{14.3.8 Example: Human Tumor Microarray Data} |
|
\bookmark[page=534,level=2]{14.3.9 Vector Quantization} |
|
\bookmark[page=535,level=2]{14.3.10 K-medoids} |
|
\bookmark[page=538,level=2]{14.3.11 Practical Issues} |
|
\bookmark[page=540,level=2]{14.3.12 Hierarchical Clustering} |
|
\bookmark[page=548,level=1]{14.4 Self-Organizing Maps} |
|
\bookmark[page=554,level=1]{14.5 Principal Components, Curves and Surfaces} |
|
\bookmark[page=554,level=2]{14.5.1 Principal Components} |
|
\bookmark[page=561,level=2]{14.5.2 Principal Curves and Surfaces} |
|
\bookmark[page=564,level=2]{14.5.3 Spectral Clustering} |
|
\bookmark[page=567,level=2]{14.5.4 Kernel Principal Components} |
|
\bookmark[page=570,level=2]{14.5.5 Sparse Principal Components} |
|
\bookmark[page=573,level=1]{14.6 Non-negative Matrix Factorization} |
|
\bookmark[page=574,level=2]{14.6.1 Archetypal Analysis} |
|
\bookmark[page=577,level=1]{14.7 Independent Component Analysis and Exploratory Projection Pursuit} |
|
\bookmark[page=578,level=2]{14.7.1 Latent Variables and Factor Analysis} |
|
\bookmark[page=580,level=2]{14.7.2 Independent Component Analysis} |
|
\bookmark[page=585,level=2]{14.7.3 Exploratory Projection Pursuit} |
|
\bookmark[page=585,level=2]{14.7.4 A Direct Approach to ICA} |
|
\bookmark[page=590,level=1]{14.8 Multidimensional Scaling} |
|
\bookmark[page=592,level=1]{14.9 Nonlinear Dimension Reduction and Local Multidimensional Scaling} |
|
\bookmark[page=596,level=1]{14.10 The Google PageRank Algorithm} |
|
\bookmark[page=598,level=1]{Bibliographic Notes} |
|
\bookmark[page=599,level=1]{Exercises} |
|
\bookmark[page=607,level=0]{15 Random Forests} |
|
\bookmark[page=607,level=1]{15.1 Introduction} |
|
\bookmark[page=607,level=1]{15.2 Definition of Random Forests} |
|
\bookmark[page=612,level=1]{15.3 Details of Random Forests} |
|
\bookmark[page=612,level=2]{15.3.1 Out of Bag Samples} |
|
\bookmark[page=613,level=2]{15.3.2 Variable Importance} |
|
\bookmark[page=615,level=2]{15.3.3 Proximity Plots} |
|
\bookmark[page=616,level=2]{15.3.4 Random Forests and Overfitting} |
|
\bookmark[page=617,level=1]{15.4 Analysis of Random Forests} |
|
\bookmark[page=617,level=2]{15.4.1 Variance and the De-Correlation Effect} |
|
\bookmark[page=620,level=2]{15.4.2 Bias} |
|
\bookmark[page=621,level=2]{15.4.3 Adaptive Nearest Neighbors} |
|
\bookmark[page=622,level=1]{Bibliographic Notes} |
|
\bookmark[page=623,level=1]{Exercises} |
|
\bookmark[page=625,level=0]{16 Ensemble Learning} |
|
\bookmark[page=625,level=1]{16.1 Introduction} |
|
\bookmark[page=627,level=1]{16.2 Boosting and Regularization Paths} |
|
\bookmark[page=627,level=2]{16.2.1 Penalized Regression} |
|
\bookmark[page=630,level=2]{16.2.2 The “Bet on Sparsity” Principle} |
|
\bookmark[page=633,level=2]{16.2.3 Regularization Paths, Over-fitting and Margins} |
|
\bookmark[page=636,level=1]{16.3 Learning Ensembles} |
|
\bookmark[page=637,level=2]{16.3.1 Learning a Good Ensemble} |
|
\bookmark[page=642,level=2]{16.3.2 Rule Ensembles} |
|
\bookmark[page=643,level=1]{Bibliographic Notes} |
|
\bookmark[page=644,level=1]{Exercises} |
|
\bookmark[page=645,level=0]{17 Undirected Graphical Models} |
|
\bookmark[page=645,level=1]{17.1 Introduction} |
|
\bookmark[page=647,level=1]{17.2 Markov Graphs and Their Properties} |
|
\bookmark[page=650,level=1]{17.3 Undirected Graphical Models for Continuous Variables} |
|
\bookmark[page=651,level=2]{17.3.1 Estimation of the Parameters when the Graph Structure is Known} |
|
\bookmark[page=655,level=2]{17.3.2 Estimation of the Graph Structure} |
|
\bookmark[page=658,level=1]{17.4 Undirected Graphical Models for Discrete Variables} |
|
\bookmark[page=659,level=2]{17.4.1 Estimation of the Parameters when the Graph Structure is Known} |
|
\bookmark[page=661,level=2]{17.4.2 Hidden Nodes} |
|
\bookmark[page=662,level=2]{17.4.3 Estimation of the Graph Structure} |
|
\bookmark[page=663,level=2]{17.4.4 Restricted Boltzmann Machines} |
|
\bookmark[page=665,level=1]{Exercises} |
|
\bookmark[page=669,level=0]{18 High-Dimensional Problems: p≫N} |
|
\bookmark[page=669,level=1]{18.1 When p is Much Bigger than N} |
|
\bookmark[page=671,level=1]{18.2 Diagonal Linear Discriminant Analysis and Nearest Shrunken Centroids} |
|
\bookmark[page=674,level=1]{18.3 Linear Classifiers with Quadratic Regularization} |
|
\bookmark[page=676,level=2]{18.3.1 Regularized Discriminant Analysis} |
|
\bookmark[page=677,level=2]{18.3.2 Logistic Regression with Quadratic Regularization} |
|
\bookmark[page=677,level=2]{18.3.3 The Support Vector Classifier} |
|
\bookmark[page=678,level=2]{18.3.4 Feature Selection} |
|
\bookmark[page=679,level=2]{18.3.5 Computational Shortcuts When p≫N} |
|
\bookmark[page=681,level=1]{18.4 Linear Classifiers with L1 Regularization} |
|
\bookmark[page=684,level=2]{18.4.1 Application of Lasso to Protein Mass Spectroscopy} |
|
\bookmark[page=686,level=2]{18.4.2 The Fused Lasso for Functional Data} |
|
\bookmark[page=688,level=1]{18.5 Classification When Features are Unavailable} |
|
\bookmark[page=688,level=2]{18.5.1 Example: String Kernels and Protein Classification} |
|
\bookmark[page=690,level=2]{18.5.2 Classification and Other Models Using Inner-Product Kernels and Pairwise Distances} |
|
\bookmark[page=692,level=2]{18.5.3 Example: Abstracts Classification} |
|
\bookmark[page=694,level=1]{18.6 High-Dimensional Regression: Supervised Principal Components} |
|
\bookmark[page=698,level=2]{18.6.1 Connection to Latent-Variable Modeling} |
|
\bookmark[page=700,level=2]{18.6.2 Relationship with Partial Least Squares} |
|
\bookmark[page=701,level=2]{18.6.3 Pre-Conditioning for Feature Selection} |
|
\bookmark[page=703,level=1]{18.7 Feature Assessment and the Multiple-Testing Problem} |
|
\bookmark[page=707,level=2]{18.7.1 The False Discovery Rate} |
|
\bookmark[page=710,level=2]{18.7.2 Asymmetric Cutpoints and the SAM Procedure} |
|
\bookmark[page=712,level=2]{18.7.3 A Bayesian Interpretation of the FDR} |
|
\bookmark[page=713,level=1]{18.8 Bibliographic Notes} |
|
\bookmark[page=714,level=1]{Exercises} |
|
\bookmark[page=719,level=0]{References} |
|
\bookmark[page=749,level=0]{Author Index} |
|
\bookmark[page=757,level=0]{Index} |
|
|
|
\end{document} |
