Driver analysis for gradient boosted trees or random forests

[BenntotheO]

Hello there,

I am looking for a method to determine the drivers or decisive variables of a gradient boosted tree model. Are there any nodes that can be used for this?(Version 5.9.0)

KG Benno

[Add94]

Highly recommend to check out the following video from knime’s channel https://youtu.be/M8wQhAylY_w?si=pOUmUiea87uSh3Oe

Around 29 mins in there’s a piece on surrogate trees and global feature importance

[rfeigel]

Try this. You can play around with the AutoML and Global Feature Importance components.

03_Global_Feature_Importance – rfeigel

[mlauber71]

@BenntotheO you can take a look at the links mentioned here. Also, there is an approach to re-enigeer the feature importance after the fact.

How to find feature importance after model building?

@somnathdas76 some model would just provide you with the feature importance (most H2O.ai models, Random Forest, XGBoost …), but you can also explore the Global Feature Importance – KNIME Community Hub Also this thread holds some more hints: I tried my hand at one idea mentioned here to re-engineer the feature importance via Tree models and the likes. But I am not 100% satisfied with the results: Then … @Morris_Kurz has provided us with two interesting nodes that also might help (TreeS…

[HansS]

Hi @BenntotheO see:

Feature Importance using decision tree

Hi @sschacht Maybe the Information Gain Calculator helps you to give some insights about the feature importance. gr. Hans