Causal Inference, Not Just Correlation To The Dependent Variable

I’ve run several analyses using Random Forest Regression, AutoML, XGBoost, GLM, and others, along with SHAP. The series of Independent Variables is associated with the Dependent Variable of Sale Price.

Aside from the frustration of each analysis arriving at disparate values for the Independent Variables (features), I need to prove some kind of causality between the IVs and DV.

I’ve spent a lot of time searching this out but have not come up with the specific KNIME Nodes or KNIME Workflows to deal with causality.

Any suggestions/links is greatly appreciated. Thanks

Hi @creedssmith - have you looked into using eXplainable AI (XAI) concepts using KNIME? This blog post links to several others that might be interesting:

We also just hosted a webinar a few weeks ago featuring @keith_mccormick that highlighted this as well. There’s a recording available here: Why AutoML needs Explainable AI (XAI) | KNIME

1 Like

Hi @ScottF thanks for your help as ALWAYS :grinning:

1 Like

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.