Question
1
Replies
163
Views
Posted: August 1, 2018
Last activity: September 3, 2018
Closed
Adaptive Modelling - Predictor Value Grouping
Hi,
Adaptive models predictors can sometime create the groups in a unexpected way, eg putting a bulk of values in a single group. I know that it is scattering it with a logic, but with the business knowledge I know that the values below shouldn't be considered in the same group. So how do I change the configuration to make sure:
-either distributing the bulk group values within other groups
-or creating new groups to do more targeted grouping
This post is somewhat similar to your other post: https://community1.pega.com/community/pega-support/question/adaptive-models-predictors-grouping-granularity and I would again emphasize that the binning is determined by statistical considerations. If you want to change segmentation (are you using models to do that?) the place for business considerations is logic.
If you talk about the binning of an individual predictor and the observation that the system automatically puts some values together in the same bin - that is done based on statistical similarity of people with these values for that attribute. There's no way to interfere with the model there (aside from rule-global settings) but you can always take specific values in consideration when consuming the output of the models.