imodels

Why we included this project

Most ML packages push you toward bigger, more complex models, but imodels goes the other way: it collects a broad set of interpretable model families, including rule lists, rule sets, decision trees, sparse linear models, and generalized additive models, all behind one scikit-learn-compatible interface. That means you can replace a black-box random forest with a concise rule list without touching the rest of your training pipeline, since the fit/predict workflow matches standard sklearn estimators. It also ships wrappers such as hierarchical shrinkage and distillation that regularize or simplify models you already have. For teams working on tabular problems where stakeholders need to understand why a prediction was made, it's a handy way to compare several transparent approaches side by side and settle on the one that balances accuracy with explainability.