Model-Agnostic Counterfactual Explanations in Credit Scoring

The past decade has shown a surge in the use and application of machine learning deep models across different domains. One such domain is credit scoring domain, where applicants are scored to assess their credit-worthiness for loan applications. During process, it key assure that there no biases or discriminations incurred. Despite proliferation (referred as black-box literature) scoring, still need explain how each prediction made by models. Most likely be prone unintended bias discrimination may occur datasets. To avoid element model discrimination, imperative during process. Our study proposes novel optimisation formulation generates sparse counterfactual via custom genetic algorithm model’s prediction. This uses publicly available Furthermore, we validated generated explanations comparing them from experts. proposed explanation technique does not only explains rejected applications, can also used approved loans.