【机器学习】6 Frequentist statistics

本章目录

6 Frequentist statistics 191
6.1 Introduction 191
6.2 Sampling distribution of an estimator 191
6.2.1 Bootstrap 192
6.2.2 Large sample theory for the MLE * 193
6.3 Frequentist decision theory 194
6.3.1 Bayes risk 195
6.3.2 Minimax risk 196
6.3.3 Admissible estimators 197
6.4 Desirable properties of estimators 200
6.4.1 Consistent estimators 200
6.4.2 Unbiased estimators 200
6.4.3 Minimum variance estimators 201
6.4.4 The bias-variance tradeoff 202
6.5 Empirical risk minimization 204
6.5.1 Regularized risk minimization 205
6.5.2 Structural risk minimization 206
6.5.3 Estimating the risk using cross validation 206
6.5.4 Upper bounding the risk using statistical learning theory * 209
6.5.5 Surrogate loss functions 210
6.6 Pathologies of frequentist statistics * 211
6.6.1 Counter-intuitive behavior of confidence intervals 212
6.6.2 p-values considered harmful 213
6.6.3 The likelihood principle 214
6.6.4 Why isn’t everyone a Bayesian? 215

github下载链接：https://github.com/916718212/Machine-Learning-A-Probabilistic-Perspective-.git