fairmodels 1.2.0 Unreleased

  • Added filtering metrics when plotting and printing of fairness_object.
  • Added ability to add custom measure function to print method of fairness_object
  • Refactored code with tidyverse style
  • Changed the order of metrics in metric_scores plot to match the ones in fairness_check
  • Added instruction for creating custom metric in README
  • Added references to vignettes
  • Enhanced the advanced vignette

fairmodels 1.1.1 Unreleased

  • Fixed error which appeared when 2 fairness objects had the same labels in them. Now if this appears it throws an error. (#41)
  • privileged parameter is now converted to character. (#41)
  • reweight() function now accepts factors (#41)

fairmodels 1.1.0 2021-05-31

fairmodels 1.0.1 2021-04-14

  • Changed examples - added parameter num.threads = 1 to ranger and added donttest{} to examples with long computation time.

fairmodels 1.0.0 Unreleased

  • Added citation information
  • Added additional reference in fairness_check() documentation.
  • Fixed links in DESCRIPTION and README.

fairmodels 0.2.6 Unreleased

  • Fixed bug which appeared when two fairness objects were passed to fairness_check without an explainer. (#36)

fairmodels 0.2.5 Unreleased

fairmodels 0.2.4 2020-12-12

  • Deleted on-load information message about four-fifths rule.
  • Fixed bug with NA warning in metrics that are not chosen. (#32)

fairmodels 0.2.3 Unreleased

  • Fixed the way the parity_loss is calculated in all_cutoffs and ceteris_paribus_cutoff. (#24)
  • Updated vignettes
  • changed documentation of functions to explicitly state metrics instead of fairness_check_metrics(). (#29)
  • Fixed typos (#27 and #28)
  • Changed conclusion drawn from density plot in Basic Tutorial (#26)

fairmodels 0.2.2 2020-09-21

  • fairness_check_data now instead of 0 has NA due to concerns of interpretability - insignificant difference could lead up to maximal value of loss. With that change when NA is created user will see warning when plotting or printing. This doesn’t affect other objects and plots.
  • Description fixes
  • Added metric_scores plot to basic tutorial
  • Updated new documentation in roc_pivot

fairmodels 0.2.1 Unreleased

  • bug related to fairness check plot fixed - rectangles did not appear for low epsilon values

fairmodels 0.2.0 Unreleased

  • adhering to four-fifths (80%) rule - changed fairness check and parity loss calculation. Now ratio is being calculated instead of differences.(#17)
  • Some plots now have default fairness metrics - same as in fairness_check
  • stack_metrics now has parameter fairness_metrics
  • corrected vignettes
  • enhanced tests

fairmodels 0.1.1 2020-08-20

  • changed examples in metric_scores function
  • changed DALEX URL in README
  • changed pre-processing to preprocessing in DESCRIPTION

fairmodels 0.1.0 Unreleased

  • main function fairness_check() implemented
  • various bias visualization functions implemented
  • pre-processing and post-processing bias mitigation techniques implemented
  • 2 vignettes present