Skip to content

Comparing fairness-aware machine learning techniques.

License

Notifications You must be signed in to change notification settings

chenhuiyeh/fairness-comparison

 
 

Repository files navigation

This repository is meant to facilitate the benchmarking of fairness aware machine learning algorithms.

The associated paper is:

A comparative study of fairness-enhancing interventions in machine learning by Sorelle A. Friedler, Carlos Scheidegger, Suresh Venkatasubramanian, Sonam Choudhary, Evan P. Hamilton, and Derek Roth. https://arxiv.org/abs/1802.04422

To install this software run:

$ pip3 install fairness

The below instructions are still in the process of being updated to work with the new pip install-able version.

To run the benchmarks:

$ from fairness.benchmark import run
$ run()

This will write out metrics for each dataset to the results/ directory.

To generate graphs and other analysis run:

$ python3 analysis.py

If you do not yet have all the packages installed, you may need to run:

$ pip install -r requirements.txt

Optional: The benchmarks rely on preprocessed versions of the datasets that have been included in the repository. If you would like to regenerate this preprocessing, run the below command before running the benchmark script:

$ python3 preprocess.py

To add new datasets or algorithms, see the instructions in the readme files in those directories.

OS-specific things

On Ubuntu

(We tested on Ubuntu 16.04, your mileage may vary)

You'll need python3-dev:

$ sudo apt-get install python3-dev

Additional analysis-specific requirements

To regenerate figures (this is messy right now. we're working on it)

Python requirements (use pip):

  • ggplot

System requirements:

R package requirements (use install.packages):

  • rmarkdown
  • stringr
  • ggplot2
  • dplyr
  • magrittr
  • corrplot
  • robust

About

Comparing fairness-aware machine learning techniques.

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages

  • HTML 84.3%
  • Python 15.4%
  • JavaScript 0.2%
  • Shell 0.1%
  • Makefile 0.0%
  • R 0.0%