What is the Fairness Tree?

The Fairness Tree is a framework developed by a group of researchers in Machine Learning and Public Policy at Carnegie Mellon University. This is a pilot version written by Lingwei Cheng.
Organizations exploring the use of Machine Learning tools need to determine relevant metrics related to bias and fairness that match their equity goals. The primary goal of the Fairness Tree is to guide users through the process of selecting the most suitable fairness criteria for their particular policy problem. It asks questions at different decision points and maps the answers to corresponding definitions of fairness. These fairness metrics ensure that individuals in one group are not discriminated against compared to other groups.

Who should use it?

The tool is particularly suitable for organizations who want to use machine learning based resource allocation tools to improve their allocation process. It is designed for machine learning/data science researchers and developers, policy practitioners who are involved in developing such tools as well as community members whose lives will be impacted by such tools.

Getting started

Starting from the top of the tree, answer each question from your perspective. You can click on each branch and read more explanation and examples about what it means to help you choose. If you encounter a question where you do not know how to choose, 1) provide feedback, elaborate on why you cannot choose; 2) after that, choose one option and continue with the flowchart until you reach an end. Go back to the question where you cannot decide, choose the alternative option, and continue with the flowchart until you reach another definition. Rank and pick which one you like better.

Provide Feedback

Please click here to provide feedback. It will take only 1 minute!

Explanation