Re: Discriminant analysis - Normality assumption

From: Jay Liu (liuj5_at_mcmaster.ca)
Date: 07/22/04 

Date: 22 Jul 2004 10:18:51 -0700

borut_rajer@hotmail.com (Borut) wrote in message news:<c67f8184.0407210132.7b89e41d@posting.google.com>...
> Hello,
>
> I need help regarding the assumptions of DA. Usually, multivariate
> normality and equality of covariance matrices are stated, but
> regarding Fisher's approach I found mentioned that the only assumption
> is the equality of covariance matrices between groups.
>
> Does multivariate normality play any role in Fisher's DA?
>
> Thank you.

As far as I know Sir Fisher did not make the normality assumption. You
can confirm this in his original paper, "The use of multiple
measurements in taxonomic problems," Annals of Eugenics, 7, Part II,
pp. 179-188, 1936.

In fact I've seen many people saying in their papers that Fisher's DA
or PCA are based on the normality assumption. This is absolutely
wrong. They should read original papers.

Relevant Pages