This preview shows page 1. Sign up to view the full content.
Unformatted text preview: onclude that b must be the eigen vector of
can conclude from (1):
( . And, we ) As the value of
is k, which is a constant, the optimal value of
would be decided by the eigen
value related to b. Thus, in order to achieve its maximal, b should be the eigen vector related to the largest eigen
value, which means
is the first discriminant variable.
(a) Page 3 of 5 We can get the conclusion that as the rank of the canonical variates increases, the centroids become less spread
out. In the lower right panel they appear to be superimposed, and the classes most confused.
(b) From this we can know that as the dimension goes larger , the training keeps reducing. But the test error reaches
its minimum at Dimension=2, then goes up again. So we can get the conclusion that in this case the best error rate
is for dimension 2.
(a) First, simulate the raw data in R ac...
View Full Document
This document was uploaded on 02/15/2014.
- Spring '14