Fairness in Data-Driven Algorithmic Loan Approval

When issuing loans, financial institutions are interested in giving loans only to customers who are likely to repay them. If a bank gives a loan to someone who repays it with interest, the bank makes a profit. However, if the loan is not repaid, the bank incurs a loss. Therefore, a bank will aim to maximize its profit.

At the same time, banks are required not to discriminate against different groups of people based on certain protected characteristics (such as gender, origin, religion).

Anyone applying for a loan wants to be evaluated independently of such protected traits or affiliations. In this learning module, you will develop and try out various strategies for distributing loans as “fairly” as possible among different groups of people. To do this, you will first explore the underlying data, develop a decision-making system for loan approval, evaluate it, and then implement various fairness strategies.