How do I restrict variables from being used more than once in a decision tree analysis?

newsasuser2017 — Tue, 26 Dec 2017 04:00:45 GMT

I am trying to run a decision tree analysis in Enterprise Miner. I have 6 variables (lets say A-F) which I am using as the inputs and have one binary target variable. When I run the analysis, I get a couple leaf nodes where the variables are repeated. For e.g. the analysis finds A as the strongest predictor and uses that as the root node and then halfway through one of the branches uses A again as the splitter variable. Is there a way I can prevent this from happening and force it to use each variable as the splitter variable only once at the maximum? I have SAS 9.4.

Re: How do I restrict variables from being used more than once in a decision tree analysis?

art297 — Tue, 26 Dec 2017 04:18:09 GMT

I don't know if the option is in EM's decision tree, but both proc split and proc hpsplit have an option that does that, namely USEVARONCE (see: http://support.sas.com/documentation/cdl/en/stathpug/67524/HTML/default/viewer.htm#stathpug_hpsplit_syntax01.htm ). You might want to check to see if it is available in EM's decision tree node.

Art, CEO, AnalystFinder.com

topic Re: How do I restrict variables from being used more than once in a decision tree analysis? in SAS Data Science

How do I restrict variables from being used more than once in a decision tree analysis?

Re: How do I restrict variables from being used more than once in a decision tree analysis?