How to train large number of models?

Selo733

Hi,

I need to train large number of models like 300 models to predict 300 different targets. I thought of using auto model to train these models since training them one by one will take too much time, months if not years in best case. But the problem is as far as I know for every model, I need to use the interface and select inputs and target, and inspect the results which will also take too much time.

Is there any way to automatize the auto model, so that these can be trained by themselves, and the best model is deployed automatically?

The second question is let's say that we managed to train these models and deployed them and after a while I noticed data drift for some models.  Is there any way to trigger a retraining process for these models? This retraining must be done by auto model too.  We can assume that the latest data is available for training.

dang

Have you take a look at the new operator for multi label modeling and its tutorial process? You would have to build multiple models manually from scratch.
https://docs.rapidminer.com/latest/studio/operators/modeling/predictive/ensembles/multi_label_model_learner.html
For your second question, we usually deploy a pre-trained model as web service on AI Hub and monitor model performance over time on a dashboard. On the model leaderboard, you can switch between active models and challenger models.