s-news
[Top] [All Lists]

interpretation of xerror in rpart

To: s-news@lists.biostat.wustl.edu
Subject: interpretation of xerror in rpart
From: michael westphal <mi_westphal@yahoo.com>
Date: Fri, 17 Mar 2006 12:09:07 -0800 (PST)
Domainkey-signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=yahoo.com; h=Message-ID:Received:Date:From:Subject:To:MIME-Version:Content-Type:Content-Transfer-Encoding; b=RwIb3pIRbbR/Wv5g+c1/7NcPtWhu+TcAIQNgyMU1IODdIhh36z4i/KeQglqbNjIENtDxRCPDy9IxflHW5aRwP6zUkt4pewdQrgUNA3JxtqSh7tNttkH1AgeSFg5qmaI/ujDgyEJ2qtdcRwcozvtJt5joSeNROKQ9KnySNrel8iM= ;
Hello:
 
I am using the rpart library in S-Plus 4.5.  I see that the correct interpretation for the Relative Error is 1- R^2.  What exactly is the interpretation of xerror, the cross-validation error?  It is related to the model's predictive ability, but numerically what does it mean? Is it possible for the xerror to be greater than 1.0?  Thank you.
 
Cheers,
 
Michael


Yahoo! Mail
Use Photomail to share photos without annoying attachments.
<Prev in Thread] Current Thread [Next in Thread>
  • interpretation of xerror in rpart, michael westphal <=