PHD Discussions Logo

Ask, Learn and Accelerate in your PhD Research

Question Icon Post Your Answer

Question Icon

2 months ago in Data Science By Rohan

Where can i find real datasets to test fairness in machine learning?

I'm building a model that's supposed to be fair across demographic groups. But to test it, I need real data where bias might actually exist. What datasets do researchers actually use?

All Answers (1 Answers In All)

By Pavitra Answered 1 month ago

A few have become standards. COMPAS is the famous one recidivism predictions with race and outcomes. Adult Census Income is great for testing bias in hiring or lending decisions. German Credit is smaller but classic for credit scoring fairness. And MEPS (Medical Expenditure Panel Survey) lets you explore disparities in healthcare access and spending. These aren't perfect they contain historical biases—but that's exactly the point. You can't fix fairness in theory. You have to test it on the messiness of real, biased data.

Your Answer