Distributionally Robust Fair Principal Components via Geodesic Descents

Reference

• Code for CFPCA and preprocessed data: https://github.com/molfat66/FairML
• Code for FairPCA: https://github.com/samirasamadi/Fair-PCA

Generate Figure 1.

• run command: $ python test.py

Generate Figure 2.

• First, generate results with .pkl format by running code: $ sh scripts/ablation_study.sh
• Then, gather results and plot: $ python functions <dataset name> --res_dir <results folder>

Generate Figure 3.

• First, generate results by running code: $ sh scripts/run_multiple_k.sh <method> <path to config> <result folder>
• Then, gather results and plot: $ python agg_multiple_k.py <dataset name> --res_dir <result folder> --img_dir <output image dir>

Generate Table 1.

• First, gererate results by running code: $ sh run_all.sh <method> <path to config> <result folder>
• Then, gather results and log: $ python agg_results.sh <dataset name/space split dataset names> --res_dir <result folder>