Know-How Exchange : Research/Metrics

Hi

Could you specify the context, because the Chi-square Automatic Interaction Detector is a huge subject and your references to Bonferroni adjustments would take a week to go over? Monotonic or not!

CHAID use in data mining is fairly well documented and as this is a marketing forum, I guess that it’s along some of these lines rather than my other subject which is life science statistics.

Bonferroni adjustment are usually used in Multivariate Statistical Process Control and Signature Analysis where the usual problems of totally wrong result are addressed by looking at Eigenfactor Detection Methods, rather than the more traditional Bonferroni adjustments, Hotelling's T-squared statistics, and the generalized variance.

I.e. looking at a predictive decision tree is very useful, but when looking for scapegoats for spending the odd million in totally the wrong direction (Actually up the wrong branch!), you should be prepared to explain why you used conventional statistical methods on a novel statistical process.

Also are you looking at Entropic models or not – entropic analysis is very useful because if you believe in statistical thermodynamics, you can chuck out a lot of economic and business modelling (Particularly deterministic stuff) as total crap.

Look forward to your clarification of the area of application.

Steve Alker
Unimax Solutions