Analyze ROC curves to determine appropriate cutoffs for each group

[Ben-Drucker]

• [X] Determine max. desired false positive rate (10% Seems appropriate; can change later)
• [ ] Add field to individual classifiers object that represents cutoff at this max. desired fpr (based on ROC curve calculations)
• [ ] Consider normalizing scores for each group