9. Regression Trees
Published Online: 11 OCT 2013
Copyright © 2014 Bee Choo Tai and David Machin.
Regression Methods for Medical Research
How to Cite
Tai, B. C. and Machin, D. (2013) Regression Trees, in Regression Methods for Medical Research, John Wiley & Sons Ltd, Oxford. doi: 10.1002/9781118721957.ch9
- Published Online: 11 OCT 2013
- Published Print: 29 NOV 2013
Print ISBN: 9781444331448
Online ISBN: 9781118721957
- continuous covariates;
- regression trees;
- tree pruning
This chapter describes the use of regression trees to help identify, among patients with a particular disease or condition, homogeneous subgroups each with an identifiable but different prognosis. The aim of establishing the groups is to tailor subsequent treatment and/or care in order to improve the ensuing outcome. The classification consists of a treelike structure beginning with a single mother or root node which is split into branches to form intermediate and/or terminal nodes referred to variously as offspring nodes, daughter or son nodes, or leaf. The root node represents all patients in the study, whereas the offspring nodes correspond to (relatively) homogeneous subgroups of patients that are identified based on specific covariate information. Details of how the tree growing process is curtailed, using a tree pruning procedure, as well as difficulties associated with the regression tree approach are included.