43 of 43 people found the following review helpful:
4.0 out of 5 stars
Flawed but well-explained, July 15, 2007
This review is from: Extending the Linear Model with R: Generalized Linear, Mixed Effects and Nonparametric Regression Models (Chapman & Hall/CRC Texts in Statistical Science) (Hardcover)
"Extending the Linear Model with R" is a "sequel" of sorts to the impressive "Linear Models with R" also written by Faraway. It assumes a basic knowledge of R (you don't have to be an expert) and a decent understanding of linear models. If you don't have that background, then I would start with the before-mentioned "Linear Models with R". If you read and understood that book, then you should be more than prepared for this one.
This book covers extensions of the linear model including Generalized Linear Models (GLM's), Mixed and Random Effects Models, Nonparametric Regression Models, Additive Models (including GAM's - Generalized Additive Models), and it contains a brief introduction to Regression Trees and Neural Networks. The biggest focus is on Generalized Linear Models. The book is fairly thorough, though not exactly comprehensive, in covering the topic of GLM's and specific commonly used GLM's. The material is very well-explained and easy to follow and they do a good job at integrating code, examples, and graphs in a way that facilitates understanding of both statistical concepts regarding GLM's and also the implementation of these concepts in R. The code is especially useful and it covers most things in R that you will need for this topic, at least those available from CRAN. The book is not very rigorous regarding theory, but that only makes the book easier to read and more practical. However, I do have one complaint regarding this section. The author spends several chapters discussing various commonly used GLM's and THEN finally gets around to defining what a GLM is and covering the basic theory. This seems backwards to me and for this reason I wouldn't read the chapters in order. Also, due to the late coverage of some of the basic theories, we don't get to see the implementation and analysis of certain sub-topics (e.g. leverage and influence) in the early examples.
Mixed and Random Effects models are second in terms of attention received. The organization is better and the explanations and code integration continue to be handled well. Nonparametric Regression and Additive Models only receive one chapter apiece, but both chapters are extremely informative and they are well-explained like the rest of the book. I was especially happy to see the coverage of GAM's (it's very short but useful) since it is a moderately recent topic (1990) and many similar books only make a brief mention of them (hey, GAM's exist) if they are mentioned at all. The chapter on Regression Trees is short, but again they make sure to cover many of the important sub-topics with clarity. The Neural Networks chapter is skimpy and you won't learn much, but it was an unexpected bonus so I can't take off points for that.
Do note that this book takes a regression approach throughout, so look elsewhere for an ANOVA perspective. The book is short with plenty of room left to talk about other topics. Thus, I would have liked to see a second part devoted to an ANOVA approach since I'm the kind of person who hates having to thumb through countless books, but they are open about the book's scope so I can't really complain.
Okay, one more complaint. I would have greatly liked to see an appendix of the R functions used throughout the book with short descriptions and references to where in the book you can find the function being discussed. R Help isn't bad, so it's not a tragic omission, but it still would have been nice.
In summary, this book is extremely useful if you plan on using extensions of linear models with R. Flaws aside, it receives my recommendation.
Help other customers find the most helpful reviews
Was this review helpful to you? Yes
No
2 of 3 people found the following review helpful:
3.0 out of 5 stars
Not so hot for teaching principles, okay for case studies, September 27, 2010
This review is from: Extending the Linear Model with R: Generalized Linear, Mixed Effects and Nonparametric Regression Models (Chapman & Hall/CRC Texts in Statistical Science) (Hardcover)
As an introduction to the general linear model this book by Julian Faraway is disappointing. Every time he introduces a new topic, the topic is quickly dismissed. I wish that he would create concise paragraphs that explain the concept he is introducing. Instead, he quickly launches into R code. I just don't find his writing style illuminating.
However, as a book of case studies, R code, and worked examples, this book is pretty good. So, learn the principles from someone else (I've been using Agresti's book on categorical data analysis and a book by Quinn & Keough) and then apply them in the R setting using the examples from Faraway.
This is coming from the perspective of a statistics graduate student. As an introduction to the subject of categorical data analysis and other linear model derived techniques I believe that it falls a bit short.
Help other customers find the most helpful reviews
Was this review helpful to you? Yes
No
