-------- Original Message -------- Subject: another cva question Date: Wed, 24 Nov 2010 21:03:32 -0500 From: Jordan Mallon <[email protected]> To: [email protected] A quick question: I'm interested in discriminating groups of landmark data, but my dataset does not fulfill the assumptions of CVA (e.g., multivariate normality of samples, equal variance-covariance matrices). Am I better off just doing a simple PCA to discriminate groups instead (knowing that this isn't necessarily the most effective approach), or should I go ahead with CVA anyways (knowing that I'm breaking the rules)? Offline replies can be sent to: [email protected] Thanks, Jordan -- Jordan Mallon Ph.D. candidate Faculty of Veterinary Medicine University of Calgary 3330 Hospital Dr. NW Calgary, AB Canada T2N 4N1 http://homepages.ucalgary.ca/~jmallon/
