Existing Features from other fairness packages

Existing Features from other fairness packages:

Fairness package:

Reporting Metrics:

1. Accuracy parity metric : (TP + TN) / (TP + FP + TN + FN)
2. Demographic parity metric : (TP + FP)
3. Equalized Odds metric : TP / (TP + FN)
4. False Negative Rate (FNR) parity metric : FN / (TP + FN)
5. False Positive Rate (FPR) parity metric : FP / (TN + FP)
6. Matthews Correlation Coefficient (MCC) parity metric : (TP × TN - FP × FN) / sqrt((TP + FP) × (TP + FN) × (TN + FP) × (TN + FN))
7. Negative Predictive Value (NPV) parity metric : TN / (TN + FN)
8. Predictive Rate Parity metric : TP / (TP + FP)
9. Proportional parity metric : (TP + FP) / (TP + FP + TN + FN
10. ROC AUC parity metric
11. Specificity parity metric : TN / (TN + FP)

aif360 package:

Debiasing Strategies：

1. Reweighing
2. Adversarial debiasing
3. Reject Option Based Classification
4. Optimized Pre-Processing
5. Disparate Impact Remover
6. Learning Fair Representations
7. Calibrated Equalized Odds Post-processing
8. Equalized Odds Post-processing
9. Meta Fair Classifier
10. Prejudice Remover

Reporting Metrics

1. Statistical Parity Difference
2. Equal Opportunity Difference
3. Average Odds Difference
4. Disparate Impact
5. Theil Index
6. Euclidean Distance
7. Mahalanobis Distance
8. Manhattan Distance
9. Accuracy

About 70 different metrics, could check the API documents