InterpretML

• Values: { explainability }
  • Explanation type: { white box Shapley value partial dependence plot sensitivity analysis }
• Categories: { model-agnostic model-specific }
• Stage: { in-processing post-processing }
• Repository: https://github.com/interpretml/interpret
• Tasks: { classification regression }
• Input data: { tabular text image }
• Licence: MIT
• Languages: { Python }
• References:
  • Nori et al. - InterpretML: A Unified Framework for Machine Learning Interpretability

The InterpretML toolkit, developed at Microsoft, can be decomposed in two major components:

1. A set of interpretable “glassbox” models
2. Techniques for explaining black box systems.

W.r.t. 1, InterpretML particularly contains a new interpretable “glassbox” model that combines Generalized Additive Models (GAMs) with machine learning techniques such as gradient boosted trees, called an Explainable Boosting Machine.

Other than this new interpretable model, the main utility of InterpretML is to unify existing explainability techniques under a single API.

Interpret-Text is an extension of InterpretML to support various text models.

Glassbox models

• Explainable Boosting Machine
• Decision tree
• Decision rule list
• Linear/logistic regression

Blackbox explainers

• SHAP kernel explainer
• SHAP tree explainer
• LIME
• Morris sensitivity analysis
• Partial dependence plots

So this package contains both model-agnostic and model-specific explainers.