09-13-2010 06:27 PM
08-11-2017 04:50 PM
The Group Processing nodes in SAS Enterprise Miner allow you to build models for a large number of groups but you are not able to see the same level of detail for each model. For example, you would not be able to see the individual decision trees fit for each (or any) level of the group variable. You could always filter the data and use a Segment Profile node to profile the fitted model after using a Filter node to subset the group(s) of interest.
If you have need to analyze a large number of groups and fit a model to each, you should consider SAS Factory Miner which allows you to fit several different models to each group. It has several advantages over the group processing approach including
* runs in parallel rather than sequentially
* allows refitting individual subgroups as needed
* will not fail fitting other groups even if fitting fails for one of the groups
I hope this helps!