Manish Nagireddy

sklearn-compat additions

2

- Added `predict_proba` for `RejectOptionClassification` in the sklearn-compatible version of AIF360. - Adjusted `GridSearch` and `ExponentiatedGradient` classes to be compatible with Fairlearn 0.7.0. Worked on as part of an internship...

Add Pareto Frontier Plot Functionality

2

Create class / function which takes in a two lists of values for separate metrics (e.g. accuracy and disparate impact ratio) and generate + display the [Pareto frontier](https://en.wikipedia.org/wiki/Pareto_front). Guidance on...

Add Equalized Odds Postprocessing to Sklearn-Compatible Version

1

Implement [Equalized Odds Postprocessing](https://aif360.readthedocs.io/en/stable/modules/generated/aif360.algorithms.postprocessing.EqOddsPostprocessing.html#aif360.algorithms.postprocessing.EqOddsPostprocessing) in the sklearn-compatible version of the toolkit. For reference, see how [Calibrated Equalized Odds Postprocessing](https://aif360.readthedocs.io/en/stable/modules/generated/aif360.sklearn.postprocessing.CalibratedEqualizedOdds.html#aif360.sklearn.postprocessing.CalibratedEqualizedOdds) is implemented currently in the sklearn-compat side of aif360.

Add Demo Notebook for Calibrated Equalized Odds Method in Sklearn-Compatible Side

1

Refer to [here](https://github.com/Trusted-AI/AIF360/tree/master/examples/sklearn) for examples of other demo notebooks.

Add ThresholdOptimizer to sklearn-compatible post-processing directory

1

Add the [ThresholdOptimizer postprocessing algorithm](https://fairlearn.org/v0.5.0/api_reference/fairlearn.postprocessing.html) from Fairlearn into the sklearn-compat version of AIF360.