ANALISIS DISKRIMINAN LINIER DAN KUADRATIK (Linear and Quadratic Discriminant Analysis)

Abstract

A classification and grouping object into groups based on several factors often approached with diskriminan analysis. Process of grouping of discriminant analysis by build discriminant function of each group and count discriminant score of each object. Based on the characteristic of data, that are multivariate normal distribution, the mean between groups and variance covariance, it is discussed in the performance of the linear discriminant and quadratic analysis. Based on the result of analysis obtained for data with multivariate normal distribution, mean between groups significantly different and variance covariance homogeneous, misclassification of linear discriminant analysis is smaller than quadratic discriminant analysis, and otherwise for data with multivariate normal distribution, mean between groups significantly different and variance covariance not homogeneous, misclassification is smaller when using quadratic discriminant analysis than linear discriminant analysis.