I am working on a system for building and maintaining 20-something predictive models. Thus far, we have built models in Miner, and the scores are updated daily. However, we now want to look at how to rig this system so that we can easilly re-run the models on newer data. This could be done maually, by updating the data sources, rerunning the models, and saving the new scorecodes. A pretty tedious process with several possibilities for manual errors. So I am considering running the models in sas code (using SAS studio\Enterprise Guide). Almost all the winning models are HP Forests, so I am pretty confident we can use the same model parameters for all models. It seems a lot faster to me to convert the miner project to sas code.
I am thinking running a macro loop through all the models, and have the program update every scorecode. This leaves me with a couple of questions.
Can I obtain the same output from EG/studio as Miner, as in a proc hp4\scorecode, and thus bypass all the manual work?
Are there any major problems to this approach?
