pypi fairlearn 0.13.0
v0.13.0

latest releases: 0.15.0, 0.14.0
13 months ago

v0.13.0

Summary

This release introduces a new method implementation, PrototypeRepresentationLearner,
which is based on the work of Zemel et al. (2013) [paper link].
You can learn more about it in the documentation.

Additionally, this release includes several improvements to existing functionality:
enhanced support for relaxed constraints and multi-dimensional input data in optimization methods,
performance improvements through vectorization, updated supported Python versions
(removed 3.8, added 3.12), first steps toward Narwhals integration,
removal of unused functionality, and cleanup of local logging initialization.


๐Ÿง  New methods


โœจ New features

  • Allowed relaxed constraints (except equalized_odds) in ThresholdOptimizer
    (PR #1453) by @taharallouche
  • Allowed multi-dimensional input data X (ndims > 2) in ThresholdOptimizer and ExponentiatedGradient
    (PR #1470) by @taharallouche

โšก Performance improvements

  • Vectorized the curve interpolation step in ThresholdOptimizer to improve performance and readability
    (PR #1448) by @taharallouche

๐Ÿ‹ Narwhals integration

  • Made the preprocessing module dataframe-agnostic through Narwhals
    (PR #1525) by @FBruzzesi
  • Made the AnnotatedMetricFunction class dataframe-agnostic through Narwhals
    (PR #1545) by @FBruzzesi

๐Ÿงน Other improvements

  • Removed support for Python 3.8 and added support for Python 3.12
    (PR #1488) by @tamaraatanasoska
  • Removed the keyword argument control_features from ErrorRate.load_data,
    since it was non-functional for the method
    (PR #1531) by @StefanieSenger
  • Removed local logging using environment variable FAIRLEARN_LOGS at initialization
    (PR #1561) by @parulgupta1004

New Contributors

Returning Contributors

Full Changelog: v0.12.0...v0.13.0

Don't miss a new fairlearn release

NewReleases is sending notifications on new releases.