FRIDAY, June 27, 2003
Time: 2:00 - 3:00 PM
Constant Hall Room 1008

Title: Canonical correspondence analysis: Some new (geometric) interpretations and computations

Dr. Laxman M. Hegde
Department of Mathematics, Frostburg State University, MD

The principal component analysis and canonical correlation analysis are well known multivariate data reduction methods in statistics. However, a method that is not so well known to statisticians is the canonical correspondence analysis. This multivariate method is extensively used for studying the species-environment relationship in ecology. The general theory behind all these methods fundamentally depends on the singular value decomposition (SVD) of the real rectangular matrices. In this talk we will review the canonical correspondence analysis method, discuss its geometry in the light of the singular value decomposition, provide some new interpretations, and discuss its computations. The talk will be beneficial to the applied scientists from ecological sciences as well.