interpretml/interpret

Health v2 · maintenance only

Adoption not part of the score

6928 stars · 787 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

InterpretML is a Python library that unifies machine learning interpretability techniques, letting users train inherently interpretable glassbox models like the Explainable Boosting Machine and explain blackbox models via global and local explanations. It provides a scikit-learn-compatible API with built-in visualizations for understanding model behavior and individual predictions.

Use cases

• explain predictions of a blackbox machine learning model
• train an interpretable model as accurate as gradient boosting
• detect bias or fairness issues in my model
• explain a model for regulatory compliance
• debug why my model made a mistake
• visualize global feature importance of a model
• build interpretable models for healthcare or finance

When to choose

• you need exact, faithful explanations rather than approximations
• you want glassbox models with accuracy competitive with random forests or XGBoost
• you work in a high-stakes domain requiring transparency and auditability
• you want a unified API for both interpretable training and blackbox explanation

When to avoid

• you only need maximum predictive accuracy with no explanation requirements
• you need deep learning interpretability for neural networks specifically
• your stack is not Python

Facets

library · maturity active

machine-learning data-visualization nlp machine-learning data-science artificial-intelligence python windows cross-platform explainable-ai interpretability xai glassbox-models blackbox-explanations explainable-boosting-machine ebm scikit-learn-compatible model-explanations fairness linux macos

Member repositories

For agents

markdown · JSON · MCP: product_card(name="interpretml/interpret")