MSEP for crossvalidation "leaving one out"

[dan12345]

HI

Could someone show me/help me how to get the MSEP (mean square error of prediction) using crossvalidation "leaving one out" for a model linear model? 

[Reeza]

I don't think there's an easy way, I think you need to write a macro to do it. Here's an example of bootstrap or Jackknife and your request would be very similar:

http://support.sas.com/kb/24/982.html

There will also be lots of papers on Lexjansen.com

David Cassell's paper is well written and has good sample code on the Jackknife method (leave one out) that will also work for you.

http://www.lexjansen.com/wuss/2009/anl/ANL-Cassell.pdf