PERFORMANCE EVALUATION OF LINEAR AND QUADRATIC DISCRIMINANT ANALYSIS MODELS DEPENDING ON DATA DIMENSIONS

Abstract

This study aims to evaluate the performance of linear discriminant analysis and quadratic discriminant analysis models on three different data sets containing 100, 500, and 1000 observations. The findings show that LDA is advantageous at low data sizes, but both models perform similarly at medium and large data sizes. It has been observed that LDA shows higher performance, especially with 1000 observations, and that large data sets increase the performance of the models.