Date: Tue, 11 Jul 2000 15:45:59 EDT
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: Hongjie Wang <Hongjiew@AOL.COM>
Subject: Re: Classification table in logistic procedure.
Content-Type: text/plain; charset="US-ASCII"
One possible reason is that the ctable options in SAS is adjusting the
coefficients for scoring. It is commonly believed that if an observation is
used to develop the model and the model is then applied to this observartion,
it artificially increases the
accuracy rate. The optimal way will be to do the following
for observation i, (i=1,...n), use obs (1,...i-1,i+1,...n) to develop the
model and then score observation i. This approach is of course costly. SAS
must have a heuristic of some sort to implement this idea.
Hope that helps.