ABSTRACT
I.0 Basic Ideas and Conventions 1
I.1 Tests of Goodness of Fit and the Brownian Bridge 5
I.2 Testing Goodness of Fit to Parametric Hypotheses 5
I.3 Regular Parameters. Minimum Distance Estimates 6
I.4 Permutation Tests 8
I.5 Estimation of Irregular Parameters 8
I.6 Stein and Empirical Bayes Estimation 10
I.7 Model Selection 11
I.8 Problems and Complements 15
I.9 Notes 20
7 TOOLS FOR ASYMPTOTIC ANALYSIS 21
7.1 Weak Convergence in Function Spaces 21
7.1.1 Stochastic Processes and Weak Convergence 21
7.1.2 Maximal Inequalities 28
7.1.3 Empirical Processes on Function Spaces 31
7.2 The Delta Method in Infinite Dimensional Space 38
7.2.1 Influence Functions. The Gaˆteaux and Fre´chet Derivatives 38
7.2.2 The Quantile Process 47
7.3 Further Expansions 51
7.3.1 The von Mises Expansion 51
7.3.2 The Hoeffding and Analysis of Variance Expansions 54
7.4 Problems and Complements 62
7.5 Notes 72
ix
8 DISTRIBUTION-FREE, UNBIASED, AND EQUIVARIANT
PROCEDURES 73
8.1 Introduction 73
8.2 Similarity and Completeness 74
8.2.1 Testing 74
8.2.2 Testing Optimality Theory 85
8.2.3 Estimation 87
8.3 Invariance, Equivariance, and Minimax Procedures 92
8.3.1 Group Models 92
8.3.2 Group Models and Decision Theory 94
8.3.3 Characterizing Invariant Tests 96
8.3.4 Characterizing Equivariant Estimates 102
8.3.5 Minimaxity for Tests: Application to Group Models 104
8.3.6 Minimax Estimation, Admissibility, and Steinian Shrinkage 107
8.4 Problems and Complements 112
8.5 Notes 123
9 INFERENCE IN SEMIPARAMETRIC MODELS 125
9.1 Estimation in Semiparametric Models 125
9.1.1 Selected Examples 125
9.1.2 Regularization. Modified Maximum Likelihood 133
9.1.3 Other Modified and Approximate Likelihoods 142
9.1.4 Sieves and Regularization 145
9.2 Asymptotics. Consistency and Asymptotic Normality 151
9.2.1 A General Consistency Criterion 152
9.2.2 Asymptotics for Selected Models 153
9.3 Efficiency in Semiparametric Models 161
9.4 Tests and Empirical Process Theory 175
9.5 Asymptotic Properties of Likelihoods. Contiguity 181
9.6 Problems and Complements 193
9.7 Notes 209
10 MONTE CARLOMETHODS 211
10.1 The Nature of Monte Carlo Methods 211
10.2 Three Basic Monte Carlo Metheds 214
10.2.1 Simple Monte Carlo 215
10.2.2 Importance Sampling 216
10.2.3 Rejective Sampling 217
10.3 The Bootstrap 219
10.3.1 Bootstrap Samples and Bias Corrections 220
10.3.2 Bootstrap Variance and Confidence Bounds 224
10.3.3 The General i.i.d. Nonparametric Bootstrap 227
10.3.4 Asymptotic Theory for the Bootstrap 230
10.3.5 Examples Where Efron’s Bootstrap Fails. Them out of n Bootstraps 235
10.4 Markov Chain Monte Carlo 237
10.4.1 The Basic MCMC Framework 237
10.4.2 Metropolis Sampling Algorithms 238
10.4.3 The Gibbs Samplers 242
10.4.4 Speed of Convergence and Efficiency of MCMC 246
10.5 Applications of MCMC to Bayesian and Frequentist Inference 250
10.6 Problems and Complements 256
10.7 Notes 263
11 NONPARAMETRIC INFERENCE FOR FUNCTIONS
OF ONE VARIABLE 265
11.1 Introduction 265
11.2 Convolution Kernel Estimates on R 266
11.2.1 Uniform Local Behavior of Kernel Density Estimates 269
11.2.2 Global Behavior of Convolution Kernel Estimates 271
11.2.3 Performance and Bandwidth Choice 272
11.2.4 Discussion of Convolution Kernel Estimates 273
11.3 Minimum Contrast Estimates: Reducing Boundary Bias 274
11.4 Regularization and Nonlinear Density Estimates 280
11.4.1 Regularization and Roughness Penalties 280
11.4.2 Sieves. Machine Learning. Log Density Estimation 281
11.4.3 Nearest Neighbor Density Estimates 284
11.5 Confidence Regions 285
11.6 Nonparametric Regression for One Covariate 287
11.6.1 Estimation Principles 287
11.6.2 Asymptotic Bias and Variance Calculations 290
11.7 Problems and Complements 297
12 PREDICTION AND MACHINE LEARNING 307
12.1 Introduction 307
12.1.1 Statistical Approaches to Modeling and Analyzing Multidimen-
sional data. Sieves 309
12.1.2 Machine Learning Approaches 313
12.1.3 Outline 315
12.2 Classification and Prediction 315
12.2.1 Multivariate Density and Regression Estimation 315
12.2.2 Bayes Rule and Nonparametric Classification 320
12.2.3 Sieve Methods 322
12.2.4 Machine Learning Approaches 324
12.3 Asymptotic Risk Criteria 333
12.3.1 Optimal Prediction in Parametric Regression Models 334
12.3.2 Optimal Rates of Convergence for Estimation and Prediction in
Nonparametric Models 337
12.3.3 The Gaussian White Noise (GWN) Model 347
12.3.4 Minimax Bounds on IMSE for Subsets of the GWN Model 349
12.3.5 Sparse Submodels 350
12.4 Oracle Inequalities 352
12.4.1 Stein’s Unbiased Risk Estimate 354
12.4.2 Oracle Inequality for Shrinkage Estimators 355
12.4.3 Oracle Inequality and Adaptive Minimax Rate for Truncated Esti-
mates 357
12.4.4 An Oracle Inequality for Classification 359
12.5 Performance and Tuning via Cross Validation 361
12.5.1 Cross Validation for Tuning Parameter Choice 362
12.5.2 Cross Validation for Measuring Performance 367
12.6 Model Selection and Dimension Reduction 367
12.6.1 A Bayesian Criterion for Model Selection 368
12.6.2 Inference after Model Selection 372
12.6.3 Dimension Reduction via Principal Component Analysis 374
12.7 Topics Briefly Touched and Current Frontiers 377
12.8 Problems and Complements 381
D APPENDIX D. SUPPLEMENTS TO TEXT 399
D.1 Probability Results 399
D.2 Supplement to Section 7.1 401
D.3 Supplement to Section 7.2 404
D.4 Supplement to Section 9.2.2 405
D.5 Supplement to Section 10.4 406
D.6 Supplement to Section 11.6 410
D.7 Supplement to Section 12.2.2 413
D.8 Problems and Complements 419
E SOLUTIONS FOR VOLUME II 423
REFERENCES 437
SUBJECT INDEX 455
AUTHOR INDEX 461