Variable Importance in AutoML Experiment

Question

We have create a classification experiment using AutoML. We can not find any output related to variable importance, propensity scores, or coefficients. Please tell us how to get this or find it?

Answer 1

Hello,

Please check on the guidance: Use the interpretability package to explain ML models & predictions in Python (preview)

https://learn.microsoft.com/en-us/azure/machine-learning/how-to-machine-learning-interpretability-aml#generate-feature-importance-values-via-remote-runs

It contains guidance about Explain the entire model behavior (global explanation), Explain an individual prediction (local explanation) and also Generate feature importance values via remote runs/ local runs

Hope this helps.

Regards,
Yutong