Date: Tue, 1 Feb 2005 08:46:59 -0500
Reply-To: "Luo, Peter" <email@example.com>
Sender: "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From: "Luo, Peter" <firstname.lastname@example.org>
Subject: How Jackknife/bootstrapping Validation Works
Content-Type: text/plain; charset="iso-8859-1"
Let's take bootstrapping as example. Suppose I've re-sampled my data 100
times and now I have a 101 data sets. Should I build one model on the
original sample and use 100 bootstrapping samples to validate the model, or
do I need to build 101 models and check the difference?
The literatures didn't give a clear clue. From what I've read, it seems that
if my purpose is to validate the model parameters, I should build 101
models; but if I just want to see how well my model performs, I only need to
build one model based on the original data and then apply the model to 100
Any help would be appreciated