Explaining Black-box Model Predictions via Two-level Nested Feature Attributions with Consistency Property

Open in new window