Ex. 12.4
Ex. 12.4
Suppose you perform a reduced-subspace linear discriminant analysis for a -group problem. You compute the canonical variables of
dimension given by , where is the matrix of discriminat coefficients, and is the dimension of .
(a) If show that
where denotes Mahalanobis distance with respect to the covariance .
(b) If , show that the same expression on the left measures the difference in Mahalanobis squared distances for the distributions projected onto the subspace spanned by .
Soln. 12.4
Consider the SVD , and write where represents the first columns and the corresponding complement. It is easy to verify that
Therefore, we have
When , the second term vanishes, and we recover (a).
When , the first term is just the Mahalanobis squared distances for the distributions projected onto the subspace spanned by .