v0.8.0#

  • Relaxed checks made on X in _validate_and_reformat_input() since that is the concern of the underlying estimator and not Fairlearn.

  • Added support for Python 3.9 and 3.10, remove support for Python 3.6

  • Added error handling in MetricFrame. Methods group_max, group_min, difference and ratio now accept errors as a parameter, which could either be raise or coerce.

  • Fixed a bug whereby passing a custom grid object to a GridSearch reduction would result in a KeyError if the column names were not ordered integers.

  • CorrelationRemover now exposes n_features_in_ and feature_names_in_.

  • Added the ACSIncome dataset and corresponding documentation.

  • Added sphinxcontrib-bibtex extension to manage citations in documentation using bibtex.

  • Added support for explicitly specifying optimization objective in ExponentiatedGradient. Added support for cost sensitive classification in ErrorRate.

  • Internal performance improvements for MetricFrame. Some results may now have a more appropriate type than object, but otherwise the only visible difference should be a substantial speed increase.

  • Added plot_model_comparison() to create scatter plots for comparing multiple models along two metrics.

  • Added adversarial mitigation approaches AdversarialFairnessClassifier and AdversarialFairnessRegressor.