Hi, folks,
I have an interesting question about Principal Component Analysis. Can we
judge the distribution of the eigenvalues of a
covariance matrix (N by N) from the covariance matrix directly (before
explicit computing the eigenvalues)? By the distribution
of eigenvalues, I mean that how many percents of the sum of all eigenvalues
are contributed by the first eigenvalue.
In other words, if given 2 covariance matrices, how can I tell which
covariance matrix has more concentrated eigenvalues?
Many thanks...
Bowen
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
