Using the High Robustness Discriminant Analysis in Classification and Predictions (A Comparative Study)

Abstract

In this study, one of the important multivariate statistical analysis methods called (discriminant analysis) was reviewed, which is used for classification and prediction through the linear discriminant function of (Fisher). However, the main problem that the study addressed is how to use the discriminant analysis when one of the basic hypotheses is violated due to the presence of outliers. The main objective was how to classify and predict when there are outliers in the data under study. Therefore, a robust and resistant method with a high breaking point was proposed to obtain accurate results in classification and prediction. The robust method was compared with the classical method, and the study concluded that the robust method has a high ability to analyze data in the presence of outliers values, as the immune function gave results with high efficiency and the classification error was very low.