Variable selection in discriminant analysis in the presence of outliers

Abstract

An important problem in discriminant analysis that has received little attention in the literature, is the effect of outliers when variable selection is the first step in the analysis. We consider two methods of dealing with this problem: the classification performance of these methods, and the classification performance if possible outliers are ignored, are compared in a simulation study. In the configurations studied, we find that none of the approaches consistently outperforms the others.