Date: Wed, 27 Feb 2008 11:58:34 -0800
Reply-To: Paige Miller <paige.miller@KODAK.COM>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: Paige Miller <paige.miller@KODAK.COM>
Subject: Re: Random sample cross Validation in Proc PLS?
Content-Type: text/plain; charset=ISO-8859-1
On Feb 27, 1:52 pm, chenm...@GMAIL.COM (Ming Chen) wrote:
> Yes. I mean PROC PLS. I also assume the random cross validation is the same
> the so called k-fold cross validation. But I am not sure because I don't see
> it from the SAS manual. If it is actually the K-fold CV, what is the reason
> for setting the seed option?
> My guess is that seed will guarantee PROC PLS generates same K different
> subsets of data for each run.
> On Wed, Feb 27, 2008 at 1:39 PM, Sigurd Hermansen <HERMA...@westat.com>
> > Ming:
> > Did you mean PROC PLS .....? I haven't used PLS cross-validation, but
> > would assume that it generates estimates for five different subsets of
> > data.
> > S
> > -----Original Message-----
> > From: owner-sa...@listserv.uga.edu [mailto:owner-sa...@listserv.uga.edu]
> > On Behalf Of Ming Chen
> > Sent: Wednesday, February 27, 2008 12:18 PM
> > To: SA...@listserv.uga.edu
> > Subject: Random sample cross Validation in Proc PLS?
> > Hi,
> > I am trying to figure out random sample cross validation in Proc PLS.
> > According to the SAS manual, The test sets are defined as "test sets can
> > be selected from the observed data randomly; this is known as random
> > sample cross validation. "
> > My sample SAS code:
> > Proc data=one method=pcr cv=random(NITER=5 seed = 2008); model y = x1
> > - x10; run;
> > Is the cross validation method same as the 5-fold Cross validation? Can
> > someone please explain it in detail?
> > Thanks!
> > Ming
The SAS docs clearly explain that NITER=5 means that SAS generates 5
random subsets to exclude. What more detail do you need?
paige\dot\miller \at\ kodak\dot\com