asmolina / ML-project-fairness-aware-classification

ML Final Group Project, Fairness-Aware Classification Approaches, Skoltech 2021

ML-project-fairness-aware-classification

ML Final Group Project, Fairness-Aware Classification Approaches, Skoltech 2021

Link to the detailed pdf-report.

Requirements

For Adaptive Sensitive Reweighting model scipydirect was used. scipydirect - a python wrapper to the DIRECT algorithm. The quickest way to install is to type:

pip install scipydirect

To install scipydirect you will need the following prerequisites: python 2.7 or 3.x, a FORTRAN compiler such as gfortran (for example, for Mac OS X: gcc), numpy, matplotlib.

Datasets

Four datasets were used: Adult Income Dataset, Bank Marketing Dataset, COMPAS dataset, KDD Census income dataset. They are also available in this repository, in Adaptive Sensitive Reweighting folder, except for KDD dataset. It could be downloaded here, census-income.data.gz file.

Structure

In the folder AdaFair there is a python implementation of the AdaFair algorithm proposed by Iosifidis & Ntoutsi

In the folder SMOTEBoost there is a python implementation of the SMOTEBoost algorithm introduced by Chawla et al.

In the folder Adaptive Sensitive Reweighting there is a python implementation of the ASR+CULEP model introduced by Krasanakis et al.