Classification trees (Q1965964)

A classification tree is a technique to investigate the relation between a set of covariates and a target variable. In this paper binary target variables which are indicators of group membership are investigated. The covariates considered are the items of the questionnaire. Each covariate defines three splits of the population. In the first step of the procedure the selection is made from all splits which have the greatest heterogeneity with respect to the target variable between the two subpopulations. Then this procedure is repeated within each of the two resulting classes. Analysis of the stability of the tree is presented.