Page 2 of 2

Re: principal components to original units

Posted: Mon Feb 22, 2010 11:52 am
by EViews Glenn
You need stationarity since using the covariance or correlation matrix of nonstationary data is problematic.