Customer Reviews

Modern Applied Statistics with S

5 star:		(13)
4 star:		(2)
3 star:		(1)
2 star:		(0)
1 star:		(1)

 

 

A worthy update. The authors are matter-of-fact and straightforward. I appreciate their terse style, the broad coverage, and the many examples. It's also good that they're starting to split programming material off into its own monograph (S Programming), making this book all the more appropriate for learning, although not quite beginning, statisticians. Their considerable contribution in software is also very much appreciated.

I do not agree at all with the reviewer who chided them for including R; I say so much the better for it. I very much hope that they will continue to do so.

Another reviewer wrote "I suspect most practicioners use S+". He should have been at the UserR! 2004 conference in Vienna this past March, with 500 or so enthusiastic R users including many from big industry (financial, pharmaceutical). And Ripley is the number-one contributor to the R Help mailing list by a long way. So it is completely appropriate that R is so prominant. Many of us appreciate open source not only for its cost ($0) but also its transparency. The reviewer should take another look at R.

As for the book, it is my data anlysis bible. It gets me started in a correct direction, with very well-explained and worked out examples, which I then adapt to my own datasets. The writing couldn't be clearer, and the references to primary sources as well as non-computational statistics texts I have found to be excellent. This is the one book to own if you are more than a beginner.

The "search inside this book" feature was not available when this review was posted. Hope it helps.

CONTENTS
Introduction
Data Manipulation
The S Language
Graphics
Univariate Statistics
Linear Statistical Models
Generalized Linear Models
Non-Linear and Smooth Regression
Tree-Based Methods
Random and Mixed Effects
Exploratory Multivariate Analysis
Classification
Survival Analysis
Time Series Analysis
Spatial Statistics
Optimization
Implementation-Specific Details
The S-PLUS GUI
Datasets, Software and Libraries
References
Index

This text is very popular and frequently cited in the statistics literature. The authors do an outstanding job of displaying modern statistical methods through the S programming language. It is an intermediate level book and certainly worthy of 5 stars.

"Modern Applied Statistics with S" provides a comprehensive introduction to the use of S in either of its implementations (S-PLUS or R). I have found it to be an indespensible resource when I use R. The authors work through a variety of major topics in statistics and demonstrate the use of S with sample code and datasets. To be clear, however, this book was not written as an introduction to statistics. You'll need to look elsewhere for such a book. Consider "Experiments in Ecology: Their Logical Design and Interpretation Using Analysis of Variance" by Underwood or "Applied Linear Statistical Models" by Neter et al. for two excellent but completely different approaches.

This book is not only THE reference on statistical techniques using S-Plus/R, but also a very good statistical reference on applied statistics. As stated by the authors on the Preface, almost all the material on the book is what is covered in the M.Sc. in Applied Statistics at the Oxford University. The book covers the most used topics in Applied Statistics currently, including Linear Models, GLM, Non-linear and Smooth Regression, Tree-based methods, Random and Mixed Effect Models, Multivariate Analaysis, Classification, Survival Analysis, Time Series Analysis and Spatial Statistics.

This is *the* book to have on S+/R. It provides excellent value for its price (indeed, any price): it is concise, broad, informative. All the same, I think it would be useful to identify intended audience for this book (in my view). First, the book is not for novices in Statistics. You'll learn how to fit generalized linear models in the language, not how and why to apply such models properly. To this end, there are plenty of specific monographies, and the majority of them use R for examples. Just to name a few, Friedman, Hastie, Tibshirani, Harrell, Faraway employ R. Also, this book assume some basic knowledge of programming. R is a more elegant language than Matlab and Thinking in R becomes very natural after some practice. But I have not seen so far a tutorial on "R as a first language". Summing up, this is a great book for undergraduates in Statistics/Engineering and up, who want a comprehensive, usable reference. My only criticism is that since 2002 there have been giant changes in the language. First, R is now the main implementation of the language, with S+ being an industry-supported variant. Second, the S4 object model is here to stay and grow, and is crying for a user-friendly introduction. Lastly, the number of packages is probably twenty times what it was in 2002. SVMs, ensemble methods, shrinkage, sparse representations *are* modern applied statistics, and are underrepresented in the book.

Still, this is a must have for any applied statistician.

After weeks struggling with poor and incomplete online documentation for the R language, this book was a breath of fresh air. The first four chapters alone, which give a broad and practical overview of the R / S / S-Plus environment, are worth the sticker price. In case it wasn't obvious at first (and it wasn't to me), R is the open-source implementation of the S language, and as the authors write in their introduction, "It is similar enough that almost all of the examples in this book can be run under R."

Venables and Ripley focus on applications of R / S for statistical analysis and visualization of data sets -- not on the theory or practice of statistics itself. As a computational biologist using R for some quick data exploration (using the Bioconductor suite), I found myself repeatedly stuck on the learning curve of R. This book gave the right amount of background at the right level.

I am often loath to buy in print what can be had for free online. But as is often the case, this hardbound reference is superior -- in form and in content -- to anything I've found on the web.

[1] The authors are well-known practicers of applied statistics, who are real experts of S and R.
[2] This book is not only an introduction to Applied Statistics, but also a handbook or guide for beginners.
[3] Anybody who is interested in Data Mining can google out Ripley's recent draft named -Statistical Data Mining-, which is freely available on website. As a complement to this book, the draft is also a good reference to beginners.

New statistics books containing R code seem to be coming out weekly. At one time I used this book a lot to get ideas on R/S programs that I needed to write. I now , seldom reference it, but go to other texts that contain R code on the topic I'm investigating. I have never liked the MASS documentation and the explanations in their book aren't much better. Many, of course will disagree with this.

It is still an important book and the only place to find good code on certain topics, i.e., nnet, poor explanations aside.
If you need sample code, though, go somewhere else first.

Still, if you only have 3 or 4 statistics books with R/S code, this should be one of them for it's breadth alone.