Montonic constraints in Xgboost with hyperparameter search

Mohammed Dataiku DSS Core Designer, Dataiku DSS ML Practitioner, Registered Posts: 39 ✭✭✭

Hi there,

I'm currently using Dataiku to train a model for a prediction task. I need to impose directionality constraints on certain columns, but the current version of xgboost in Dataiku doesn't support these kinds of constraints. I've learned that we can write custom Python models with monotonic constraints as detailed here. However, these models have limitations because we can't perform hyperparameter tuning on them. I'm wondering if it's possible to expand the capabilities of these models to search for the best hyperparameters. Has anyone tried this before?

Operating system used: Windows X

Setup Info
      Help me…