Message-ID: <4937E014.3070009@vanderbilt.edu>
Date: 2008-12-04T13:50:12Z
From: Frank E Harrell Jr
Subject: Getting Odds ratios for tree branches-CART model
In-Reply-To: <f7b9e9a90812040156s655f735fve2d2f4c3c9c50a43@mail.gmail.com>
jestinah mahachie wrote:
> Dear all,
>
>
>
>> I have been using rpart to get trees. May someone help me with how I can do
>> some logistic regression modelling to derive odds ratios on the branches.
>> Also help on how to derive estimates on prediction error.
>>
>> --
>> Kind Regards,
>>
>> Jestinah
The odds ratios from rpart will be badly biased away from 1.0.
Unbiasing them will be tricky.
Frank
>>
>>
>>
>> Home Address:
>> Brusselsepoortstraat 85
>> 9000
>> Gent
>> Belgium
>>
>
>
>
--
Frank E Harrell Jr Professor and Chair School of Medicine
Department of Biostatistics Vanderbilt University