Date: Fri, 17 Dec 2004 15:27:49 -0800
Reply-To: Nico Peruzzi <firstname.lastname@example.org>
Sender: "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From: Nico Peruzzi <email@example.com>
Subject: Comparing Risk Statistics between training and testing sample in
Content-Type: text/plain; charset=US-ASCII
Thansk to all for answering all my questions today - you can tell that
I'm in the heat of things.
Using Answer Tree, doing E-CHAID with a continuous target variable.
I have a large sample so I am partitioning out a portion to test the
model. First question: what proportion of the sample does one hold
out? What are some guidelines?
When comparing the risk statistics between the training and testing
models, what does one look for?
Thanks again for all the help, Nico
Nico Peruzzi, Ph.D.