Classification Statistics Z* and Z

Abstract

Suppose we have an observation x from one of two p-variate populations NW! , Ell) and N Ell) where the parameters , and Ell are unspecified but is positive definite. Given independent random samples from these populations, the problem of classification of x into its relevant population can be tackled by using the discriminant function Z proposed by Kudo (5) and John (3, 4) and studied by Memon (6), Memon and Okamoto (8). Sometimes there occur situations in texanomical problems when in addition to the knowledge of discriminators, information is available on a covariate y whose mean is known to be the same in both multivariate populations and 7t2, that is, has population 7ti: N