s-news
[Top] [All Lists]

tree model goodness of fit/general tree model questions

To: s-news@lists.biostat.wustl.edu
Subject: tree model goodness of fit/general tree model questions
From: Stephanie A Mather <STEPHANIE.MATHER@huskymail.uconn.edu>
Date: Thu, 26 Oct 2006 09:30:57 -0400
Hello all - 
I am using S-Plus 7.0 for windows to develop a regression tree as a means to 
stratify my predictor variables in a way so that they best predict my 
continuous response variable for my dissertation.  I have little 
S-Plus/Statistics background, so I apologize in advance.  Here are a few 
questions I have:

1.  Is there some value on the summary output that gives goodness-of-fit of the 
output tree model?

2.  What exactly is deviance/residual mean deviance with respect to tree 
models?  From what I understand, res. mean deviance is the squared difference 
between observed and predicted values, which would mean a lower res. mean 
deviance is a better model, is this correct?

3.  I understand nodes split where the y-values are statistically different - 
is this at 95% confidence?  Can this be adjusted?

4.  Is there any way to compare tree models to determine which is best to use?

Any help/suggestions would be very very greatly appreciated - thank you in 
advance!

-Stephanie


Stephanie Mather
Graduate Research Assistant
University of Connecticut
Dept of Civil & Enviro. Engineering
261 Glenbrook Rd, Unit 2037
Storrs, CT 06269-2037


<Prev in Thread] Current Thread [Next in Thread>
  • tree model goodness of fit/general tree model questions, Stephanie A Mather <=