Providing a decision support system to assess the credit risk of real bank customers

Subject Areas : Financial Knowledge of Securities Analysis

Sirous Azizollahi ¹ , Mahdi Madanchi Zaj ^{2 *} , Ghasem Mohseni ³ , Mehrdad Hosseini Shakib ⁴

1 - Department of Accounting, Karaj Branch, Islamic Azad University, Karaj, Iran
2 - Department of Financial Management, Faculty of management, Electronic Branch, Islamic Azad University, Tehran, Iran ( corresponding author).
3 - Department of Accounting, Karaj Branch, Islamic Azad University, Karaj, Iran
4 - Department of Industrial Management, Karaj Branch, Islamic Azad University, Karaj, Iran

Received: 2023-08-27 Accepted : 2023-10-24 Published : 2025-01-25

Keywords: Real customers, Credit risk, Credit Scoring, Multiple logistic regression, Decision Support System,

Abstract :

The purpose of this research is to provide a decision support system to assess the credit risk of real bank customers. For this purpose, firstly, using the library method, the effective indicators on the credit risk assessment of real customers were identified and then, using the fuzzy Delphi method, the indicators were screened. The community of this section was formed by banking experts (Bank Mellat) who were selected by the snowball method. In the following, The decision support system was designed and implemented to evaluate the credit risk of real bank customers in SPSS Modeler software. The final data related to the indicators included the files of 7318 real customers of Bank Mellat during the years 2013-2016.The results of the research showed that the decision support system consisting of support vector machine and random forest correctly predicted the status of customers in the four categories of on-time delivery, past due, overdue and questionable delivery by 67.09% and 65.10%, respectively. Meanwhile, the combined support vector machine-random forest model has provided a better prediction with a rate of 77.17%.