ce-mondrian-conditional

What problem does it solve?

Configure and validate Mondrian (conditional) calibration to expose subgroup-specific uncertainty estimates for fairness-aware deployments in calibrated explanations.

Core Features & Use Cases

• Mondrian-based conditional calibration that partitions calibration data by group to produce per-bin uncertainty intervals.
• Supports three bin specification options: inline bins, MondrianCategorizer for continuous features, and a callable mc.
• Provides a clear evaluation workflow and references examples in references/mondrian_examples.md for hands-on guidance.

Quick Start

Load your dataset, choose a Mondrian binning approach (Inline bins, MondrianCategorizer, or mc callable), calibrate, and validate subgroup-specific uncertainty estimates.