As any data scientist will tell you, training and evaluation of "fair" classifiers is challenging, in part because most fairness metrics of interest depend on both the "sensitive" attribute information and the label information of the data points. Much of this is a trial-and-error exercise. TruEra assists you with this by letting you change the fairness metric applied for a preview of protected segment scores before drilling down into specific segment comparisons.
Comments
0 comments
Please sign in to leave a comment.