Designing an Interpretable Credit Scoring System Using Fuzzy Association Rules

“Credit Scoring (CS)” refers to methods for evaluating the creditworthiness of individuals or legal entities applying for bank loans. The credit rating is typically expressed as a numerical score, with higher scores indicating more creditworthy loan applicants. However, this process carries severe risks in financial management. For this reason, Credit Risk(CR) analysis has become an important topic and, recently, one of the primary goals of the banking and financial sectors. The use of credit risk analysis for personal and business bank credits has become very common. This paper employs fuzzy association rule mining to extract frequent rules for good and bad credit applicants from a dataset. The extracted rules were applied to the CS system and correctly classified 79.9% of the credit requests. A key advantage of this method is that it provides an interpretable model capable of classifying requests with adequate accuracy.”