Representative selection of variables

Abstract

The classical and most frequent use of the discriminant function is for the purpose of classification, by obtaining a combined measurement such that the difference between two or more groups is as large as possible. The linear discriminant function for the special case of two groups was first discussed by R. A. Fisher (see [7]). A systematic generalization of the concept of discriminant functions was presented by S. N. Roy (see [11]), who utilized the discriminant function for the purpose of constructing tests of multivariate hypotheses. Discrimination between two groups has been widely utilized and is now discussed in many standard textbooks.