Credit Risk Assessment Using Machine Learning in Banking Sector

The possibility of financial loss as a result of a borrower’s failure to fulfil their financial obligations is referred to as credit risk. Even though there are many elements that affect credit risk, rigorous evaluation before loan approval (credit scoring) and continued observation of client payments and behavior can help reduce the risk of non-performing assets (NPAs) and fraud. The necessity for reliable methods of projecting loan performance has been highlighted by a noticeable rise in NPAs and fraud cases over the past few years. Researchers are looking at various ML techniques to increase the accuracy of credit risk assessment in light of the increasing expansion of artificial intelligence (AI), especially in machine learning (ML), driven by increased internet access, data availability, and computer capacity. In this model, we used Logistic Regression, Support Vector Machine, Random Forest and XGB algorithms to get the most accurate result. While customer profiles from credit rating and scoring firms are paid for by banks, continuing research is focused on utilizing ML to improve credit risk assessment. The present paper reviews 15–20 credit risk evaluation studies that were published between 2012 and March 2023. The analysis not only looks at the present limits of these studies but also identifies new research trends in the area.