Skip to contents

Model-agnostic explainability (XAI) for bias discovery

Source: R/fairness_xai.R
morie_fairness_xai.Rd

R ports of the explainer suite in morie.fairness.xai. Permutation importance, partial dependence, ALE, ceteris-paribus profiles and SHAP-style attributions are computed natively in base R from first principles (iml is not used). Every callable takes a predict_fn closure (matrix -> numeric vector) so it works on any classifier or risk model.

Details

  • morie_fairness_xai_permutation_importance

  • morie_fairness_xai_partial_dependence

  • morie_fairness_xai_ale

  • morie_fairness_xai_ceteris_paribus

  • morie_fairness_xai_shap_values