You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
# Debiasing Representations by Removing Unwanted Variation Due to Protected Attributes
Here we present an example using the ProPublica COMPAS data set. Using race as the protected attribute, we compare raw and debiased predictions trained on a simple logistic regression model. We show that the debiased data decreases the magnitude of the difference of both false positve and false negative rates between African-Americans and Caucasians will little loss to accuracy. While our model is simple, we also show it is comparable to the COMPAS algorithm in scoring and accuracy.