Developing a Novel Fair-Loan Classifier through a Multi-Sensitive Debiasing Pipeline: DualFair

Machine learning (ML) models are increasingly being used for high-stake applications that can greatly impact people’s lives. Sometimes, these be biased toward certain social groups on the basis of race, gender, or ethnicity. Many prior works have attempted to mitigate this “model discrimination” by updating training data (pre-processing), altering model process (in-processing), manipulating output (post-processing). However, more work done in extending situation intersectional fairness, where we consider multiple sensitive parameters (e.g., race) and options black white), thus allowing greater real-world usability. Prior fairness has also suffered from an accuracy–fairness trade-off prevents both accuracy high. Moreover, previous literature not clearly presented holistic metrics with fairness. In paper, address all three problems (a) creating a bias mitigation technique called DualFair (b) developing new metric (i.e., AWI, measure algorithm based upon inconsistent counterfactual predictions) handle Lastly, test our novel method using comprehensive U.S. mortgage lending dataset show classifier, fair loan predictor, obtains relatively high metrics.