Presenting the Smart Pattern of Credit Risk of the Real Bank Customers Using Machine Learning Algorithm

In the past, loan approval decisions for bank customers in Iran were traditionally made based on personal judgments regarding the risk of repayment. However, the increased demand for banking services from economic enterprises and families, coupled with heightened and extended competition among banks and financial institutions in the country to reduce facility repayment risk, has necessitated the adoption of novel methods, including statistical approaches. Today, bankers em-ploy customer credit ranking to predict the risk of default in banking facility re-payment and classify candidates. This new approach offers several advantages, including time efficiency, cost-effectiveness, elimination of personal judgments, and enhanced precision when assessing applicants seeking various forms of funding. Numerous statistical methods, including bias analysis, logistic regres-sion, non-parametric parallelism, as well as other techniques such as neural net-works, have been applied to credit ranking. In this study, a smart model for real bank customer credit risk, based on the random forest metaheuristic algorithm, is presented, with a focus on the case study of Bank Tejarat. Based on the skewness value, the data can be considered to exhibit a normal distribution. The results reveal that the variable related to the type of facility had the lowest mean, while the maximum value was associated with the facility amount.