Recursive Partitioning in the Health Sciences (Statistics for Biology and Health)

Rating:
Summary: sequel to CART
Review: Brieman, Olshen, Friedman and Stone introduced CART in their 1984 book. It is an effective methodology and software tool for constructin classification and regression trees. The procedure is also referred to as recursive partitioning. There has been a great deal of research over the past 16 on this topic and the authors cover the basics and the new material well. New ideas include survival trees and adaptive splines (including MARS). It provides interesting applications to health science problems. Th authors compare tree based methods to logistic regression. This is a notable successor to the CART text.

Rating:
Summary: sequel to CART
Review: Brieman, Olshen, Friedman and Stone introduced CART in their 1984 book. It is an effective methodology and software tool for constructin classification and regression trees. The procedure is also referred to as recursive partitioning. There has been a great deal of research over the past 16 on this topic and the authors cover the basics and the new material well. New ideas include survival trees and adaptive splines (including MARS). It provides interesting applications to health science problems. Th authors compare tree based methods to logistic regression. This is a notable successor to the CART text.

Rating:
Summary: Recursive Partitioning
Review: Recursive Partitioning in the Health Sciences is one of the few statistical texts specifically written with the epidemiologist as a target end user, similar in genre to Schlesselman's Case Control Studies. The subject matter is relatively new in the field of epidemiology and as such needs to be related contextually to more traditional statistical approaches. The authors accomplish this by incorporating introductory chapters on methods corresponding to those being addressed by the nonparametric methods of recursive partitioning and multivariate adaptive regression splines (MARS). Additionally, they compare results between these tried and true statistical methods and recursive partitioning and MARS with many illustrative examples. This last is a strength of this book. Examples of each topic under discussion are carefully considered in a stepwise manner. The book is nicely balanced in terms of theoretic background and practical applications, with the writing generally intelligible to the non-statistician. The book has provided our group with background material to allow utilization of recursive partitioning in our research. As the technique of recursive partitioning becomes recognized and subsequently applied in the epidemiological field, this book may well become a classic.

Rating:
Summary: Recursive Partitioning in the Health Sciences
Review: Zhang and Singer have done a splendid job of explaining recursive partitioning, a topic that should be of great interest to anyone who wants to make sense of data in which there are many potentially important variables contributing to some outcome or variable of interest. One should not be put off by the "... in the Health Sciences" part of the book's title; the potential audience of readers who can benefit from reading it is much greater than this implies (I'm an ecologist, for example). Why? First, because the topics covered have wide applicability in many fields; and second, because the writing is exceptionally clear and easy to follow. If you are able to use a typical introductory text on multiple regression, for example, you should have no difficulty getting a lot out of Zhang and Singer. If you are able to handle a mathematically rigorous approach to statistics but are new to the topics covered here, this book will provide an excellent starting place before you jump into the many references to the recent literature provided by the authors.