Seminar Details

Seminar Details


Oct 15

3:30 pm

Interpretation of Eigenvalues in Multivariate Statistical Analysis

Steen Andersson


Indiana University

In this paper, we define general canonical correlations and the corre-
sponding general covariate pairs, which generalize the classical canonical correlations and covariate pairs developed by Hotelling. We also define so-called canonical variance distances and the corresponding canonical distance variates. In a generalsetting, these parameters and their corresponding empirical statistics are interpreted and also characterized in terms of certain eigenvalues and eigenvectors. In particular, these concepts are also applied to the ten fundamental testing problem
discussed by Andersson, Brøns, and Tolver Jensen (1983)