Classification and discriminant analysis

• Main Author: Fazeli, Goldisse
• Format: Others
• Published: 2000
• Online Access: http://spectrum.library.concordia.ca/1085/1/MQ47800.pdf; Fazeli, Goldisse <http://spectrum.library.concordia.ca/view/creators/Fazeli=3AGoldisse=3A=3A.html> (2000) Classification and discriminant analysis. Masters thesis, Concordia University.

This study provides a comprehensive review of the literature pertaining to the problem of classification. General concepts and principles of the classification problem are explored. These results are presented especially for populations under a normal distribution. Three major techniques of classification and discriminant analysis are presented: linear discriminant analysis, quadratic discriminant procedures and logistic regression. Logistic regression is reviewed in its general framework and as a classification tool. A few articles on the comparison of the efficiency of discriminant analysis and logistic regression are summarized. The discriminant approach is proven to be more efficient in the case of populations with a multivariate normal distribution. Under nonormality, logistic regression with maximum likelihood estimators outperforms discriminant analysis.