Using a CHAID model on a flag target, enabling boosting to enhance model accuracy (while all other parameters are set to their defaults) results in the following execution error: "Warning: Error: Field $R1-Target has an incorrect storage type in the data model."
Model runs fine without boosting and $R1-Target seems like an internally calculated field ... so I don't understand what's going on.
Answer by Mauricio Cornejo (479) | Feb 04, 2016 at 02:32 PM
I raised an issue with Technical Support. Turns out it is a bug when CHAID is used from auto-classifier. There is an interim fix.
I'm running Modeler 17 with Fix Pack 1 (i.e. 17.0.0.1) and the link I was sent for applying IF020 is
For anyone that thinks they are hitting this issue, log a ticket with technical support. We can help you verify you are experiencing the defect, then give you access to the fix Mauricio noted above.
Answer by Kenneth Jensen (IBM) (2577) | Feb 03, 2016 at 12:05 PM
This does not look like an error or warning that I have seen before and I think that you are correct in your suspicion that the $R1-Target
is an internally calculated field; probably related to the first of the component models estimated as a part of the boosting.
I would suggest that you contact our Technical Support to open a PMR for them to investigate this error.