AI GitHub
InterpretML for explainable predictive models
Fit explainable models and inspect feature effects with shared visualization tools.
Key Takeaways
- InterpretML includes Explainable Boosting Machines and tools for exploring contributions.
- Decide whether to fit an interpretable model or inspect an existing one.
- Fit an EBM and compare its accuracy with a black-box baseline.
- Feature contributions explain the model's behavior.
How it works
InterpretML includes Explainable Boosting Machines and tools for exploring contributions. Global and local views answer different questions about the same model.
What you need
Decide whether to fit an interpretable model or inspect an existing one. Both need clear feature names and an evaluation set drawn from the intended use.
What to try first
Fit an EBM and compare its accuracy with a black-box baseline. Inspect a feature curve and a prediction for an unusual example.
Inspect the number of training examples behind a surprising feature effect. Curves in sparse regions can be less reliable than patterns supported by many observations.
Where to look for failures
Feature contributions explain the model's behavior. Correlated inputs and data collection choices can still produce misleading real-world interpretations.
Related Franklin pages
An explanation of a prediction and an estimate of a causal effect ask different questions. Inspect the data and assumptions required by the selected analysis before interpreting its output.
- Attribute a prediction to input features using explainer methods suited to the model. SHAP for explaining model predictions
- Estimate causal effects under a stated model of relationships and assumptions. DoWhy for explicit causal analysis
For a primer on learning from data and evaluating predictions, see Franklin’s glossary. Machine learning
Project documentation
Try it for your workflow
Explore the project’s documentation for setup and usage, or save this page to a collection for later.
Read the project documentation ↗