EVALUATION OF THREE CLASSIFICATION RULES FOR MIXTURE OF DISCRETE AND CONTINOUS VARIABLES

Abstract

The best classification rule is the one that leads to the smallest probability of misclassification which is called the error rate. This work focused on three classification rules for mixture of discrete and continuous variables with the aim to evaluate the performance of these rules to in classification of individuals into several categories. Applications were done using simulated data and real life data. The result obtained revealed that the location model achieved better result than the other two rules in minimizing the average error rate in both datasets.

Keyword: Location Model, Linear Discriminant Models, Quadratic, Discriminant Model, Error Rate.