Recursive partitioning in the health sciences信息编码、析取与分配的数学

1999-3  

Springer Verlag  

Zhang, Heping/ Singer, Burton  

226  

This book describes the recursive partitioning methodology and demonstrates its effectiveness as a response to the challenge of analyzing and interpreting multiple complex pathways to many illnesses, diseases, and ultimately death. For comparison purposes, standard regression methods are presented briefly and they are applied in the examples. We emphasize particularly the importance of scientific judgment and interpretation while guided by statistical output. This book is suitable for three broad groups of readers: 1) Biomedical researchers, clinicians, public health practitioners including epidemiologists, health service researchers, environmental policy advisers; 2) Consulting statisticians who can use the recursive partitioning technique as a guide in providing effective and insightful solutions to clients' problems; and 3) Statisticians interested in methodological and theoretical issues. The book provides an up-to-date summary of the methodological and theoretical underpinnings of recursive partitioning. It also presents a host of unsolved problems whose solutions whould advance the rigorous underpinnings of statistics in general. Heping Zhang is Associate Professor of Biostatistics and Child Study at Yale University. In addition to the methodology and application of recursive partitioning, he is interested in developing statistical methods for analyzing correlated data, especially family and genetic studies, and brain imaging problems. Burton Singer, a member of the National Academy of Sciences, is Professor of Demography and Public Affairs at Princeton University. His research interests include combinatorial formulation of randomness, infectious disease epidemiology, and bio-demography of aging.

Preface1　Introduction　1.1 Examples Using CART　1.2 The Statistical Problem　1.3 Outline of the Methodology2 A Practical Guide to Tree Construction　2.1 The Elements of Tree Construction　2.2 Splitting a Node　2.3 Terminal Nodes　2.4 Download and Use of Software3　Logistic Regression　3.1 Logistic Regression Models　3.2 A Logistic Regression Analysis 4　Classification Trees for a Binary Response　4.1 Node Impurity　4.2 Determination of Terminal Nodes　 4.2.1 Misclassification Cost　 4.2.2 Cost Complexity　 4.2.3 Nested Optimal Subtrees　4.3 The Standard Error of Rcu　4.4 Tree-Based Analysis of the Yale Pregnancy Outcome Study　4.5 An Alternative Pruning Approach　4.6 Localized Cross-Validation　4.7 Comparison Between Tree-Based and Logistic Regression Analyses　4.8 Missing Data　 4.8.1 Missings Together Approach　 4.8.2 Surrogate Splits　4.9 Tree Stability　4.10 Implementation5　Risk-Factor Analysis Using Tree-Based Stratification　5.1 Background　5.2 The Analysis6　Analysis of Censored Data: Examples　6.1 Introduction　6.2 Tree-Based Analysis for the Western Collaborative Group Study Data7　Analysis of Censored Data:Concepts and Classical Methods　7.1 The Basics of Survival Analysis　 7.1.1 Kaplan-Meier Curve　 7.1.2 Log-Rank Test　7.2 Parametric Regression for Censored Data　 7.2.1 Linear Regression with Censored Data　 7.2.2 Cox Proportional Hazard Regression　 7.2.3 Reanalysis of the Western Collaborative Group Study Data8　Analysis of Censored Data: Survival Trees　8.1 Splitting Criteria　 8.1.1 Gordon and Olshen's Rule　 8.1.2 Maximizing the Difference　 8.1.3 Use of Likelihood Functions　 8.1.4 A Straightforward Extension　8.2 Pruning a Survival Tree　8.3 Implementation　8.4 Survival Trees for the Western Collaborative Group Study Data9 Regression Trees adn Adaptive Splines for a Continuous Response10 Analysis of Longitudinal Data11 Analysis of Multiple Discrete Responses12 AppendixReferencesIndex

Recursive partitioning in the health sciences信息编码、析取与分配的数学 PDF格式下载