IBM Support

Model built with 'large datasets' option does not calculate propensity.

Technote (troubleshooting)


Problem(Abstract)

Working with C&R Tree, have selected in model's Build Options; Objectives "Create a model for very large datasets" and in the Model tab have selected "Calculate raw propensity scores."

Once the model is built, do not see a $RP-response field; no propensity is generated.
Make change in Objectives, select Boosting or Bagging and then there is a $RP-response field.
Why is the model not creating a propensity field when created through the 'Create a model for very large datasets option'?

Resolving the problem

This problem has been reported to SPSS Development - Workaround Available

Workaround:
Use the prediction and the confidence to calcuate the raw propensity -- if prediction = 1 propensity = confidence, if prediction = 0 propensity - 1 - confidence.

Historical Number

92948

Document information

More support for: SPSS Modeler
Modeler

Software version: 14.0

Operating system(s): Windows

Reference #: 1489389

Modified date: 05 December 2012