You are here

The Elements of Statistical Learning: Data Mining, Inference, and Prediction

Trevor Hastie, Robert Tibshirani, and Jerome Friedman
Publisher: 
Springer
Publication Date: 
2009
Number of Pages: 
745
Format: 
Hardcover
Edition: 
2
Series: 
Springer Series in Statistics
Price: 
89.95
ISBN: 
9780387848570
Category: 
Textbook
[Reviewed by
Peter Rabinovitch
, on
05/28/2012
]

We are fortunate that there are now several excellent texts for machine learning. Some are based on particular software systems, like Python, Matlab, or R, and some focus more on the math, putting the software in the background. In the latter category, three of my favorites are the relatively elementary Introduction to Machine Learning by Ethem Alpaydin, the more advanced, and more Bayesian, Pattern Recognition and Machine Learning by Chris Bishop and the even more advanced but less Bayesian book being reviewed here.

Machine learning is a large, rapidly expanding field, and the book under review is the second edition — the first was only in 2001. This new edition has several chapters on topics not in the original: random forests, ensemble methods, undirected graphical models, and high dimensional problems (the so-called “wide” data). There are also many minor changes throughout the text to what was already a great book.

The book would be ideal for statistics graduate students who have had a course at the level of, say, Casella & Berger’s Statistical Inference. It would be hard to work through the complete book in one semester, but the instructor could pick and choose topics, and then the student will have the book as a reference. This book really is the standard in the field, referenced in most papers and books on the subject, and it is easy to see why.

The book is very well written, with informative graphics on almost every other page. It looks great and inviting. You can flip the book open to any page, read a sentence or two and be hooked for the next hour or so.

Each chapter ends with bibliographic notes and exercises, and most have a “computational considerations” section too.

The complete book is available, legally, from the authors’ web site. If you are at all interested in machine learning, I’d strongly suggest you download it and have a look. I suspect that you will soon thereafter purchase it to have in a more convenient format.


Peter Rabinovitch is a Systems Architect at Research in Motion. He just defended his PhD thesis on Mallows permutations, and therefore is very tired.