Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!usc!zaphod.mps.ohio-state.edu!pacific.mps.ohio-state.edu!linac!midway!msuinfo!learning.siemens.com!scheuche
From: scheuche@pleiades.cps.msu.edu (Anton Scheucher)
Newsgroups: comp.ai
Subject: ID3 and Chi-square test
Keywords: ID3, Chi-square test
Message-ID: <1991Jan16.195516.21123@msuinfo.cl.msu.edu>
Date: 16 Jan 91 19:55:16 GMT
Sender: news@msuinfo.cl.msu.edu
Reply-To: scheuche@pleiades.cps.msu.edu (Anton Scheucher)
Organization: Michigan State University
Lines: 24

HI, 

I have a short question regarding the ID3 learning algorithm by 
J.R. Quinlan. In his paper 'Induction of Decision Trees' he 
describes how a Chi-square test can be used to 'prevent testing any 
attribute whose irrelevance cannot be rejected with a very high
confidence level'. 

-1- Is this Chi-square test part of the ID3 algorithm or already a 
    modification of it?
-2- Is this test usually used?
-3- Does this test effect the performance of ID3, when only a small 
    number of instances is used (about 10 - 20)? (I am aware of the
    fact that the Chi-square test should NOT be used for small 
    numbers.) 

Thanks,

Anton Harald

----------------------------------------------------------------
Scheucher Anton
Michigan State University   email: scheuche@pleiades.cps.msu.edu
----------------------------------------------------------------