06-19-2017 03:59 PM
I have built a SAS Miner project. The diagram has 20 models and the model comparator is choosing the best model at run time. The idea is, in the long run, when more data comes in, the models will be refreshed automatically and automatically the best model will be picked up and scored.
How can this end to end automation be achieved? In model manager it seems only the best model coming out of the model comparator is picked up and that too a static version of the model - with the derived variables and coefficient of latest model training. Is there a way to deploy the entire diagram in model manager (or SAS DI maybe), so that evertime I trigger or schedule a run, all the models are refreshed using new data and new model variables and coefficient are selected and the best model is selected for scoring - all at run time?
Any suggestion will be greatly appreciated.