I'm using the TS model factory tool since it allows me to run the model against multiple groups but realize that it doesn't work with the TS compare tool. Are there any alternatives so that I can compare both models across multiple groups and determine which performs best against out-of-sample data?
Hi @sheidari ,
You might need to wrap the groups into a batch macro and run them through the standard tools to allow you to correctly use the TS Compare.
Isolate the values of the control parameter (what would determine in a filter the group number/name) then feed the data in and run it for each group.
Give it a group output value so you can then split and compare.
Alternatively, you could run them through the compare tool in an iterative macro and only keep the best fit model on each run. Kind of like Survival Analysis.
M.
User | Count |
---|---|
17 | |
15 | |
15 | |
8 | |
6 |