rules

The rules package is a parsnip extension that provides model definitions for rule-based machine learning models within the tidymodels framework. It enables fitting and prediction with three types of rule-based models: cubist models with linear rules and ensemble boosting, classification rules derived from decision trees, and rule-fit models that combine tree-extracted rules with regularized regression.

This package solves the problem of integrating rule-based models into tidymodels workflows using a consistent interface. It supports both classification and regression tasks across multiple engines (C5.0, Cubist, and xrf). Rule-based models offer interpretable predictions through human-readable if-then rules while maintaining competitive predictive performance compared to black-box models.