Second Edition (Springer Series in Statistics) - and Prediction

I would strongly dissuade anyone from buying this book. Approximately 50% of the book is mathematical notation and the author minimally defines the notation that he employs. This is a great book for someone that understands the notation used, but if you are not well versed in symbols then this might as well be Greek. The mathematics itself is not difficult, but the terms that he uses require substantial on-line research. The author also changes the meanings for certain symbols without informing the reader. To me, it appears that he is almost purposefully attempting to obfuscate the material. He definitively targets mathematical PhD students at the expense of other types of students. I am finishing up a Masters in Analytics and this was the worst book that we read thus far. Multi-factor and logistic regression are fairly straightforward concepts and it is laughable how the author overcomplicates such benign material. I question the author's ability to actually apply these concepts in a practical business setting.

These types of books will cease to exist in a few years as new ones emerge that actually explain and teach these concepts in a lucid manner. I am actually speaking with a publisher about publishing such a book. If you truly wish to understand these concepts AVOID THIS TOME!

There is no denying that this book is widely beloved, held in high regard, and referred to as the bible for machine learning.

If you are interested in learning Machine Learning or wish to strengthen your understanding you might be tempted to acquire this book. However, I will urge you to postpone such an impulse after considering the following:

In the introduction of The Elements of Statistical Learning (ESL), it has a section stated "Who should read this book", which explains who the target audience is and what the authors expect the reader (you) to know. It states:

We expect that the reader will have had at least one elementary course in statistics, covering basic topics including linear regression

This line deserves admonishment. Unless your elementary course in statistics either taught or had the prerequisites of linear algebra from an abstract perspective, multivariate calculus (through a linear algebra perspective , e.g. taking matrix derivatives), probability, and functional analysis you - the reader - will likely be very lost in this book.

This book assumes you know a lot.
It skips over derivations and will simply state properties (sometimes with the assumption that the reader will know why they are stated).
It also lacks any sort of mathematical prelude / appendix to remind the reader of concepts they should be familiar with prior to reading this book.
If you wish to take part in some of the problems at the end of each chapter, you should also be familiar with proofs.

These critiques are only fair in so far of the statement from the authors in regard to their expectations of the reader (that you have had at least one elementary course in statistics).

The authors have put a lot of care into trying to condense a swath of material into a concise reference. A book providing all of the prerequisite material would be massive and this book's goal is not to teach the fundamentals of mathematics; rather teach the elements of statistical learning. For achieving that goal the book has earned the praise it is due.

So in short, if you wish to learn about machine learning and have a loose mathematical background you are better of not starting here. Go build up your fundamentals. A surface read of this book will leave you no better off than having not read it.

If you have had some more courses than that single elementary course in statistics and are armed with the community of Math StackExchange, you can fight your way through this book (although it won't be pretty).

For purebred mathematicians, you should be fine.

Lots of people consider this the standard reference on classical ML. Maybe it depends on the fact that I’ve a financial and economics background, maybe the text is really too dense and messy, but I’ve a different opinion. From time to time crucial facts are condensed in a one phrase observation, so you risk to miss many important things and don’t appreciate the relative importance of the different techniques encyclopedically presented. A more complete, up to date and easier way to approach this proteiform matter is to study the two MIT texts: Murphy (Machine Learning: A Probabilistic Perspective) and Goodfellow-Bengio-Courville (Deep Learning).
Purchased though the store.it

Very comprehensive reference text. I don't understand why people enjoy it so much though. The book doesn't present statistics in an interesting way at all. I have only read chapter 7 so far (on model selection and assessment) and the material is presented like a list of definition after definition. Model selection is such an important and controversial topic both in industry as well as research and it disappoints me that the authors take such a dry and neutral stance. They make some of the most important aspects of statistics seem excruciatingly boring.

Data mining is a field developed by computer scientists but many of its crucial elements are imbedded in important and subtle statistical concepts. Statisticians can play an important role in the development of this field but as was the case with artificial intelligence, expert systems and neural networks the statistical research community has been slow to respond. Hastie, Tibshirani and Friedman are changing this.
Friedman has been a major player in pattern recognition of high dimensional data, in tree classification, regularized discriminant analysis and multivariate adaptive regression splines. He has also done some exciting new research on boosting methods.

Hastie and Tibshirani invented additive models which are very general types of regression models. Tibshirani invented the lasso method and is a leader among the researchers on bootstrap. Hastie invented principal curves and surfaces.

These tools and the expertise of these authors make them naturals to contribute to advances in data mining. They come with great expertise and see data mining from the statistical perspective. They see it as part of a more general process of statistical learning from data.

The book is well written and illustrated with many pretty color graphs and figures. Color adds a dimension in pattern recognition and the authors exploit it in this book. It is really the first of its kind that treats data mining from a statistical perspective and is so comprehensive and up-to-date.

The important statistical tools that are covered in this book include under the category of supervised learning; regression, discriminant analysis, kernel methods, model assessment and selection, bootstrapping, maximum likelihood and Bayesian inference, additive models, classification and regression trees, multivariate adaptive regression splines, boosting, regularization methods, nearest neighbor classification, k means clustering algorithms and neural networks. These methods are illustrated using real problems.

Similarly under the category of unsupervised learning, clustering and association are covered. They cover the latest developments in principal components and principal curves, multidimensional scaling, factor analysis and projection pursuit.

This book is innovative and fresh. It is an important contribution that will become a classic. The level is between intermediate and advanced. Good for an advanced special topics course for graduate students in statistics. A comparable text is the text by Mannila, Hand and Smyth.

This book made effective use of color and maintained a competitive price. This had a major impact on publishers like Wiley that could not sell a book at this size and initial price. Wiley is still looking for a book comparable to this one that they can use to compete with Springer-Verlag. I know this information because I heard from the Wiley acquisitions editor that I worked with on my two books.

This book is used by many machine learning courses. It is used in the Stanford grad program, which should give everyone enough understand of the authors targeted audience. Do not expect to sit down and just read it like a novel for a quick overview of statistical learning methods. Warning, expect some heavy duty math. In the interest of full disclosure I'll repeat: expect mega-math. The authors claim you can read the book and avoid what they term "technically challenging" sections, but I'm not really sure how one would do that. The book presents just about every important ML technique from decision trees to neural nets and boosting to ensemble methods. The Bayesian neural nets are tons of fun.

You can download a pdf copy from the authors website to take a look at it, but a serious student in the subject really should get hardcopy. [...]

I have three texts in machine learning (Duda et. al, Bishop, and this one), and I can unequivocally say that, in my judgement, if you're looking to learn the key concepts of machine learning, this one is by far the worst of the three. Quite simply, it reads almost as a research monologue, only with less explanation and far less coherence. There's little/no attempt to demystify concepts to the newcomer, and the exposition is all over the map. There simply isn't a clear, coherent path that the authors set out to go on in writing a given chapter of this text; it's as if they tried to squeeze every bit of information of the most recent results into the chapter, with little regard to what such a decision might do to the overall readability of the text and the newcomer's understanding. To people who might disagree with me on this point, I'd recommend reading a chapter in Bishop's text and comparing it to similar content in this one, and I think you'll at least better appreciate my viewpoint, if not agree with it.

So you might be wondering, why do I even own the text given my opinion? Well, two reasons: (1) it cost 25 dollars through Springer and a contract they have with my university (definitely look into this before buying on the store!), and (2) if you actually already know the concepts, it is quite useful as a summary of what's out there. So to those who understand the basics of machine learning, and also have exposure to greedy algorithms, convex optimization, wavelets, and some other often-utilized methods in the text, this makes for a pretty good reference.

The authors are definitely very well-known researchers in the field, who in particular have written some good papers on a variety of machine learning topics (l1-norm penalized regression, analysis of boosting, to name just two), and thus this book naturally will attract some buzz. It may be very useful to someone like myself who is already familiar with much of what's in the book, or someone who is an expert in the field and just uses it as a quick reference. As a pedagogical tool, however, I think it's pretty much a disaster, and feel compelled to write this as to prevent the typical buyer -- who undoubtedly is buying it to learn and not to use as a reference -- from wasting a lot of money on the wrong text.

An outstanding book! Several folks on here are complaining that the book is difficult to follow because of the mathematics. Well, I hate to break it to those folks, but machine learning IS hard! There’s a reason why the best data scientists and predictive modelers are mathematicians and statisticians.

To my knowledge, this was the first comprehensive human-readable textbook on the topic. If you're proficient in Greek, you may want to give "A ProbabilisticThoery of Pattern Recognition" a try. I'm always referring to The Elements of Statistical Learning for nuggets of insight into how statistical learning actually works.

You can download the book for free from the authors' websites or (I believe) the publisher's site.

Content: Amazing. This is a must read for anyone that wants to do data science or machine learning.

Quality: Here's where the book fails. This is a math book, so I expect the math content to look good. I'm not exactly sure what the process involved in making a kindle book is, but the images of formulas are terrible. The images are really fuzzy and some even cut off part of the text. That being said, I ended up returning the book and getting elsewhere due to this.

The book is very wordy and reads like a newspaper or a novel. A simple one line equation, or an elementary concept,
takes many pages to explain. The result is 745 pages of never-ending text. The least square and maximum likelihood
methods are presented over and over with different notations.

I bought it because the authors were professors at Stanford. But the book is not good.

Really tough to follow the flow of points, most details are hidden or kept as exercises. The concepts would be referred from before and after the current section accidently, making a coherent understanding difficult. Any unfamiliar term easily added by the authors potentially incurs cognitive overload, as well as frustration for those readers without such backgrounds. Unluckily, those commented 'easy' points are quite difficult to me.
Much better experience for me when I was reading MLAPP and Deep Learning.

That's what happened to me respect this book. Conclusion: DO NOT BUY ANY BOOK, EVER, WITHOUT FLIPPING IT. OTHER'S REVIEW ARE NOT RELIABLE. This is the third time I bougth a book based on other's review. I repented.

This book is a miracle of clarity and comprehensiveness. It presents a unified approach to state of the art machine learning techniques from a statistical perspective. The layout is logical and the level of math is appropriate for applications-oriented engineers and computer scientists, as well as theorists. Sections where the book does need to go into heavier mathematics are clearly marked and generally optional. I found the book very easy to read, but at the same time very comprehensive.
The book provides a very illuminating counterpoint to other books that promote the Computational Learning Theory (COLT / kernels / large margins) viewpoint of modern machine learning. Many of the same techniques such as boosting and support vector machines are discussed, but are motivated in different ways. Appropriate regularization is seen as the key to avoiding overfitting with complex models, rather than margin maximization. Mathematically you often end up solving the same problem, but personally I usually find the statistical approach much more direct and intuitive.
This book is a nice follow on to introductory pattern recognition texts such as Duda and Hart, though it can be read without any prior pattern recognition knowledge. It provides a nice mix of theory and paractical algorithms, illustrated with numerous examples. An essential element of your machine learning library!

This book describes most of the important topics in machine learning. Most machine learning books just present a criterion and and an optimization algorithm. For instance, LDA is often presented as: here is the Fisher criterion, it seems like a good thing to maximize. "The Elements of Statistical Learning" also presents that this is the right criterion if the distributions of the data for each class are Gaussian with the same covariance. This book puts all the algorithms in the same statistical language, which makes them easy to compare and choose between.

I also appreciate the emphasis this book puts on algorithms that are more recently popular/effective. I very much appreciate the discussions of logistic regression vs. LDA, ridge and lasso regression, boosting/additive logistic regression and additive trees, decision and regression trees, ...

The only qualm I have with this book is that it is rather biased toward the authors' own research. It is difficult from reading this book alone to differentiate between classical techniques and the authors' recent proposed algorithms.

i really like this book. i haven't finished reading yet. it's extremely dense. by that, i mean every page, every paragraph is packed full of information. it makes for slow but very rewarding reading. i bought the book because

i wanted to learn something about the topic. i've got a math and statistics background, but i haven't dealt with the broad topic of data mining or statistical learning. the book suits my needs very very well.

it's clearly written. i haven't found any grammatical or technical errors. it's pacing is ambitious, but i find i can follow it. i do think some math and statistics background is required to make the book readable and useful.

i wouldn't hesitate to recommend it to someone with the appropriate background.

This is an excellent book, but is probably most accessible to readers who already have a pretty solid grounding in statistics or applied mathematics. I have used it as background material for courses in advanced data analysis and computational statistics for PhD students in management, but I would generally not use at as a textbook for these students, because their typically mathematical preparation is too patchy. I do use it as a text for research assistants who work with me closely and develop a level of mathematical maturity.

That being said, this is a very comprehensive overview of the field, and is incredibly useful as a reference for statisticians and other professionals.

The review from September 8 expresses an opinion which is the exact opposite of mine, and is worded so strongly that I have to object. I gave a course using the book to bioinformaticians, most of them with a computer science background, and found the book exceptionally well prepared and suitable for a graduate course. The book serves the dual purpose of an introduction and a reference. An especially nice feature is how the authors explain the relationships and differences between different methods. By doing so, they provide context which I have not seen in any other book on this subject. The book is a very nice combination of basic theory and performance evaluation on data from a wide variety of domains and it is quite up-to-date. It has a well developed website going with it and the graphical material can be obtained electronically from the publisher. The book is an outstanding contribution to the field.

These guys have made a great contribution to the statistical literature. It is a broad book that entends to summarize the latest methods available for data analysis. The authors succeed in giving a statistical context with which to compare and contrast many statistical methods. Some of the statistical methods discussed were developed in the past 5-15 years (SVM, boosting, LASSO, etc...) and haven't yet been put into a broader context. While this book is not comprehensive in its treatment, it is the best single book on data analysis available.

Among my commercial data mining friends this book is considered to be the bible. It is worth having just to assess the mindset of the day-to-day data miners. The book discusses many data mining issues in more depth than most of the earlier works on this subject. However it still lacks the the depth and counsel of, say, applied multiple regression books (cf. Draper and Smith) that give guidance on when a particular method may give false results or how bogus results can be detected posteriori.

Definitely this book is awesome. But I have only one concern about this kind of book ( where mathematical equations are there in the book), kindle versions are difficult to read. Fonts of mathematical equations are not good in terms of readability on Kindle, even if I increased the size of font, it doesn't increase the fonts in the equation. So finally I had to return this Kindle edition purchase.

If you are looking for a relatively rigorous but very readable data mining book, this is simply the best! It covers most of the modern techniques and is beautifully printed with high quality graphics.

The book is comprehensive no doubt however there so many loose ends in the explaining of concepts and techniques that you feel the book would have been much better pruned down a hundred pages or so. The exercises are also not really clearly defined and sometimes you feel that you have to do mind reading just to understand what they are asking for. Further on the notation in the book even though revised is a jumble and things keep showing up out of the blue without explanation or reference. The authors sometimes also take overly complicated mathematical paths to solutions rather than the intuitive and leave out the motivation which leads to a lot of frustration.(For example the Fisher Discriminant or the EM algorithm).

To give an example of the overly complicated way of writing look at this sentence on minimising intra-cluster point distance from the cluster mean:

"Thus, the criterion is minimized by assigning the N observations to the K clusters in such a way that within each cluster the average dissimilarity of the observations from the cluster mean, as defined by the points in that cluster, is minimized."

The book is full of these passive voice and 40 word sentences. It would be awesome if the authors read: Style: Ten Lessons in Clarity & Grace and then did a complete overhaul of the book.

This is not an introduction to statistical learning theory. It is a collection of overviews of various statistical methods presented rather than explained to the reader. In order to benefit from this book the reader should have a good background in matrix algebra and should already have a theoretical and working knowledge of the topics covered. For detail on the methods and their real world application the reader should also be prepared to consult other references. Two stars because, fairly or not, it does not have the pedagogical value that I expected of it.

Very entertaining and in-depth review of the topic. But the topic is a lot of different things and there seems to be a bit of a mismatch between the content of the book, the title, and the the store categories it is given. Data mining, inference, and predeiction of course, probably have *something* to do with artificial life, but thats not the first thing a reader experts to read about for this kind of topic.

I did enjoy it but expectation management is key. It just ended up being about something a bit different than expected.

I was a good quantitative treatment of several different issues. It could have done a better job of explaining why that particular set of issues was a contiguous group of ideas. I could have imagined them talking about several different concepts as well.

The graphics are great. More stats books should spread their wings with some interest-keeping color.

This book is a very interesting book to learn the main statistical approach of data mining. It's clear and full of examples. If you go a Stanford data mining website you will find all the courses and exercises linked to the book.

An important book to have in your own data mining library.

It gives a complete overview and middle-depth discussions on a wide thematic statistics. Additionally provides methodological elements for making decisions on the implementation of specific techniques. Very good book. I'm an economist and statistical and I was very useful.

In 2009, the second edition of the book added new chapters on random forests, ensemble learning, undirected graphical models, and high dimensional problems. And now, thanks to an agreement between the authors and the publisher, a PDF version of the 2nd edition is now available for free download. [...]

Model averaging is an integral part of model selection and prediction.

Bayesian model averaging (BMA) is a highly sophisticated method for doing model averaging.

It is amazing then that the book hardly touches upon BMA or Bayesian methods.

(By my counting there is a total of 7 pages in all).

In my opinion the authors are very weak in this area which

explains why the topic of BMA is not covered. This is a

shame because it is more than likely BMA would be a serious

competitor (if not better) than the other methods they are

familiar with and discuss in the text.

This book has become a classic for any statistician and data miner by now.

It is a broad overview of regression, classification and clustering techniques (supervised and unsupervised machine learning).

Desde el punto de vista tecnico este libro es sumamente completo ya resume muy bien muchos de los metodos estadistico que se aplican en la mineria de datos. Pero a mi parecer este libro es debil en cuanto a la descripcion de ciertos algortimos que describe, y en algunos casos solo se menciona el procedimiento como una formula (la matriz S de los trazadores cubicos) ya que presupone que el lector ya tiene un buen conocimiento sobre trazadores cubicos.

Este no es un libro introductorio de mineria de datos, se tiene que tener un cierto nivel en analisis estadistico multivariado

I owned this book. This book introduces machine learning from the traditional stat point of view. I would recommend other books because (1) You don't want to read on if you find so many typos in the first several pages; (2) The data used in the book is still the iris data, which has only 150 points. How many data Google has to process?