Hi there, I was just wondering if there is a node to extract which features were important in XGBoost Tree model building for regression? Or are there other nodes I would need to reconfigure to use ti get the feature importance? Thank you

Not possible with nodes but with python scripting. See: [modkzs] How to get feature importance in xgboost? python, xgboost asked by modkzs on 08:05AM - 04 Jun 16 UTC And official xgboost documentation: https://xgboost.readthedocs.io/en/latest/python/python_api.html?highlight=get_score#xgboost.Booster.get_score Note that it’s important to see that xgboost has different types of “feature importance”. The Solution: What is mentioned in the Stackoverflow reply, you could use SHAP to determine feature importance and that would actually be available in KNIME (I think it’s still in the KNIME Labs category). Therefore if SHAP suit your purpose (and it should), that is the way I would go.

XGBoost Feature Importance

KNIME Analytics Platform

ipazin August 17, 2021, 1:08pm 3

Hello @Subha_D,

see this topic as well:

Br,
Ivan

2 Likes