Credit Approval Scoring Using Ensemble Learning

One of the most delicate problems in the banking industry is credit risk assessment. Manually assessing a client's credit risk is a time-consuming and laborious task. As a result, an automated model for the financial sectors must be built that increases prediction accuracy and decreases application access time. This paper presents an innovative credit scoring model. For class balancing in the considered dataset, it uses the robust oversampling technique for class balancing. The Bagging approach combines base learners to create the ensemble learning based model. The Australian creditworthiness dataset is used to test applicability. The suggested model achieves the highest results with values of 89.42%, 89.02%, and 95.27% for accuracy, F-measure, and AVC, respectively.