We are currently in the process of creating Version 1.0-beta, to be released in 2021. If you would like to be notified of this release, please send me an email!
This tutorial provides an introduction to the importance of analyzing machine learning predictors for fairness. It is recommended (but not required) that you are familiar with python and working with pandas dataframes before starting.
In this tutorial, you will learn:
- How to identify training data that might have redundantly encoded bias
- An introduction to how fairness models can be used in supervised learning
- How to fix a basic discriminatory predictor with a fairness model through:
- Understanding perfmorance metrics
- Minimizing loss
- Utilizing sensitivity, recall, and precision
It is recommended to have Jupyter Notebooks installed on your machine. However, you can run a Notebook in the browser.
There are two tutorial files. One with solutions and one without. Simply clone or fork the repo to access the tutorial Jupyter Notebook on your personal machine.
This tutorial was created with an accompanying paper, that has been submitted to the Journal of Statistics Education. The paper is primarily to help teachers and educators understand the background of Fairness Models in the Machine Learning before teaching the contents of the tutorial to students.
Until the paper is publicly available, please reference the Tutorial Solution Manual to check your answers and to get help on the tutorial.
If you would like to add content to this tutorial, check out the Contributions Guidelines to see what I'm looking for. In general, I would love additional tutorials with solution manuals and reference material for fairness models.
If you'd like to reach out to discuss this work or colloborate, please send me an email!
jessiejsmith01@gmail.com
This Tutorial is Open Source under the Creative Commons License 4.0.
