From unbiased MDI Feature Importance to Explainable AI for Trees

Abstract

We attempt to give a unifying view of the various recent attempts to (i) improve the interpretability of tree-based models and (ii) debias the the default variable-importance measure in random Forests, Gini importance. In particular, we demonstrate a common thread among the out-of-bag based bias correction methods and their connection to local explanation for trees. In addition, we point out a bias caused by the inclusion of inbag data in the newly developed explainable AI for trees algorithms.

arXiv admin note: text overlap with arXiv:2003.02106

References in corpus (4)

• Scikit-learn: Machine Learning in Python
• A Debiased MDI Feature Importance Measure for Random Forests
• Unbiased variable importance for random forests
• Data-driven advice for interpreting local and global model predictions in bioinformatics problems

Cited by in corpus (1)

• Data-driven advice for interpreting local and global model predictions in bioinformatics problems