A Review of the Limitations of Some Discriminant Analysis Procedures in Multi-Group Classification.

Abstract

A review is given on existing work and result of the performance of some discriminant analysis procedures under varying conditions. Few of the developed methods (Fisher’s Linear Discriminant Function, Logistic Regression and Quadratic discriminant function) were reviewed. Some new results are presented for the case involving allocation with more than two groups. Shortfalls in the reviewed procedures necessitated the need for an improved procedure that can classify observations into multiple groups with high efficiency (minimal error rate).

Keywords: Multivariate, Discrimnant function, Classification, Multi-groups, Optimal