CHAID decision tree to determine significant cutoff of continuous variable

[Jai Sule]

I am using Stata 13 SE and have installed the chaid and chaidforest packages. I have a dataset of patients with ct aortic calcifications labelled ctcal and change in operative strategy labelled opstratch as categorical variables and Age as a continuous variable. I would like to use the chaid decision tree algorithm to determine the significant Age cutoff among patients with ctcal=1(yes) who have had change in op strategy opstratch=1(yes). What syntax should I type in to be able to do this?

[Joseph Luchman]

Hi Jai,

I would recommend looking over the helpfile included with the software and the examples included therein.

It is not clear that what you are seeking to do with -chaid- (or -chaidforest-) is something that is obtainable with the software.

-chaidforest- is a predictive algorithm primarily and it is unlikely to be of much use for what you are seeking to do without using other decision tools based on the predicted values it produces.

You may be able to obtain information about some cut points from the -chaid- decision rules but the very specific need noted here may not be simple to accommodate with what -chaid- provides. This is why I recommend taking a look to see what to expect in terms of results.

- joe