The second edition of the classic textbook introduction to statistical learning has come. New contents such as in-depth learning are available for free download
Heart of machine 2021-08-08 20:44:31
Stanford classic textbook 《The Element of Statistical Learning》( abbreviation ESL) Statistical learning known as the frequency school 「 Bible 」, from Trevor Hastie、Robert Tibshirani、Jerome Friedman The three masters worked together to complete . This book introduces neural network Support vector machine 、 Classification tree and boosting、 Graph model 、 Random forest, etc machine learning Algorithm , It can help readers understand machine learning Algorithm overview .

But because this book involves a lot of formulas 、 Matrix derivation and many other difficult contents , It is more suitable for individuals with advanced training in Mathematical Science , And for beginners , It's difficult to learn , therefore ,Trevor Hastie They wrote another entry-level book 《Introduction to Statistical Learning with R( Introduction to statistical learning : be based on R application )》(ISL), Help more people get started as soon as possible .

ISL Weakens the details of mathematical derivation , Pay more attention to the application of methods , amount to ESL Guide to reading , Very popular among entry-level readers . Each chapter contains an experiment , use R Language implementation . These experiments provide readers with valuable practical experience .

at present , The book has been translated into many languages , Including Chinese 、 Italian 、 Japanese 、 Korean 、 Russian and Vietnamese .

Now? ,ISL Updated to the second edition (ISLRv2), Compared to the first edition , The second edition adds Deep learning ( The first 10 Chapter )、 Survival analysis ( The first 11 Chapter )、 Multiple tests ( The first 13 Chapter ).ISLRv2 Some chapters of the first edition have also been greatly expanded : Naive Bayes And generalized linear model ( The first 4 Chapter ), Bayesian additivity Back to the tree ( The first 8 Chapter ), Matrix completion ( The first 12 Chapter ).

Besides ,ISLRv2 The whole... Has been updated R Code experiments .

Book address :

Netizens strongly recommend : very nice , This is a great introductory book .

Book Introduction

There are as many books as 597 page , common 13 Chapters , It mainly includes :

  • Statistical learning ;

  • Linear regression ;

  • classification ;

  • Resampling Method ;

  • linear Model selection And Regularization ;

  • Nonlinear models ;

  • Tree based approach ;

  • Support vector machine ;

  • Deep learning ;

  • Survival analysis and deletion data ;

  • Unsupervised learning ;

  • Multiple tests .

To be specific , This book 2 This chapter introduces the basic terms and concepts behind statistical learning , In addition, it introduces k - Nearest neighbor classifier , This is a very simple method , Very effective in dealing with many problems .

The first 3 Zhang He 4 This chapter introduces the classical linear methods for regression and classification . In particular , The first 3 Chapter reviews Linear regression , This is the basic starting point of all regression methods ; The first 4 Chapter discusses two of the most important classical classification methods , Logistic regression and linear discriminant analysis .

A core problem in all statistical learning situations is to choose the best method for a given application . therefore , The first 5 This chapter introduces cross validation and bootstrap, They can be used to estimate the accuracy of many different methods to select the best method .

The first 6 Many linear methods are considered in this chapter , Including classical and more modern linear methods , They provide a reference to the standard Linear regression Potential improvements to , Including ridge regression 、 Principal component regression and Lasso etc. .

The remaining chapters are mainly nonlinear statistical learning . The first 7 First, some nonlinear methods are introduced , These methods can solve the problem of only one input variable , Then it shows how to use these methods to fit a nonlinear additive model with multiple inputs .

The first 8 In this chapter, the tree based method is studied , Include bagging、boosting And random forest .

The first 9 Describes the Support vector machine The content such as .

The first 10 Describes the Deep learning , This is a kind of non Linear regression And the method of classification , In recent years, it has received wide attention .

The first 11 This chapter discusses survival analysis , This is a regression method , It is specially used when the output variable is deleted , That is, incomplete observation .

The first 12 This chapter introduces the unsupervised setup , There are input variables in the unsupervised setting , But no output variables . Specially , Principal component analysis is proposed 、k - Mean clustering and hierarchical clustering . Last , The first 13 This chapter discusses the very important topic of multiple hypothesis testing .

The authors introduce

The authors of the books are... From left to right :Gareth James、Daniela Witten、Trevor Hastie and Rob Tibshirani.

Gareth James Is the associate dean of the Marshall School of business at the University of Southern California , The main research fields include functional data analysis 、 High dimensional regression 、 Statistical problems in marketing .

Daniela Witten Is an American biostatistician , The main research area is how to use machine learning To understand high-dimensional data .

Trevor Hastie I'm a professor at Stanford University , It used to be AT&T A technician at Bell Labs .2018 year ,Hastie He was elected to the National Academy of Sciences . His main research field is applied statistics .

Rob Tibshirani He is a professor in the Department of statistical and biomedical data science at Stanford University , He has developed statistical tools for analyzing complex data sets .

Book catalogue :

本文为[Heart of machine]所创,转载请带上原文链接,感谢