A Model Classification Technique for Linear Discriminant Analysis for Two Groups

Linear discriminant analysis introduced by Fisher is a known dimension reduction and classification approach that has received much attention in the statistical literature. Most researchers have focused attention on its deficiencies. As such different versions of classification procedures have been introduced for various applications. In this paper, we attempt not to robustify the Fisher linear discriminant analysis but to propose a comparable model for dimension reduction and classification. The proposed model is investigated and compared with Nearest mean classifier and Fisher classification rule using unscaled normal and scaled normal generated data. Numerical simulations reveal that the proposed model performed exactly as Fisher’s approach and outperformed nearest mean classifier.