A NEW METHOD OF MULTIPLE DISCRIMINANT ANALYSIS
A new method of multiple discriminant analysis was developed in this paper. At first, the hyperellipsoidal approximate domains of the G normal populations were found by using the idea of principal component analysis based on the variance-covariance matrices of G samples from the known normal populations,In addition,the euclidean distance was defined as an assistant discriminant index. Then, the classification of theunknown samples was discussed based on the hyperellipsoidal approximate domains and the assitant discriminant index. In the end,the method was used in some examples and the efficiency of the method and the problems associated with the efficiency were described.