Seminarsselected review on unified framework of discriminant analysis
148
reads
Starting with the pioneering work of Fisher (1936), discriminant analysis has become one of the most important techniques in applied statistics. The classical linear discriminant function introduced by Fisher achieves this separation by maximizing the standardized difference between the means of these two populations. On the other hand, a logistic regression with linear combination of covariates is also used. In this talk, we will present some attempts in the literature to put those approaches into a unified framework, In particular, we will address the connection between the Neyman-Pearson Lemma, ROC curve and AUC. Part of the talk is based on Eguchi and Copas (2002, 2005, 2006)
reads
Hung Chen
2008-05-27
13:30:00 - 15:00:00
405 , Mathematics Research Center Building (ori. New Math. Bldg.)
Starting with the pioneering work of Fisher (1936), discriminant analysis has become one of the most important techniques in applied statistics. The classical linear discriminant function introduced by Fisher achieves this separation by maximizing the standardized difference between the means of these two populations. On the other hand, a logistic regression with linear combination of covariates is also used. In this talk, we will present some attempts in the literature to put those approaches into a unified framework, In particular, we will address the connection between the Neyman-Pearson Lemma, ROC curve and AUC. Part of the talk is based on Eguchi and Copas (2002, 2005, 2006)