Local MDI+ computes sample-specific feature importances for random forests and boosted trees by combining tree split structure with regularized linear models, and it outperforms LIME, TreeSHAP, and Local MDI at identifying predictive features.
MDI regresses only on in-bag samplesΨ(X∗;S)
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Local MDI+: Local Feature Importances for Tree-Based Models
Local MDI+ computes sample-specific feature importances for random forests and boosted trees by combining tree split structure with regularized linear models, and it outperforms LIME, TreeSHAP, and Local MDI at identifying predictive features.