freediscovery.metrics.
ratio_duplicates_score
Given cluster labels x and y, compute the relative error between the number of duplicates in x vs the one in y.