Document Type : Article
Department of Financial Engineering, Faculty of Industrial Engineering & Systems, Tarbiat Modares University, Jalal AleAhmad Ave., Tehran, Iran. P.O.Box: 14115-111
Department of Financial Engineering, Faculty of Financial Science, Khatam University, Hakim Azam Street, Tehran, Iran
In developed countries, the existence of credit scoring agencies helps to reduce the credit risk of banks across the globe by ensuring their good credit scores through a variety of techniques, including the use of machine learning and artificial intelligence. Nevertheless, the banking system in poor and developing countries is plagued by a lack of the reputable agencies for credit scoring of customers. As a result, their banks tend to internalize scoring according to Basel II & III and their Central Bank regulations. In this study, eight machine learning techniques were used to rank the credit scores of legal customers of an Iranian bank. The optimal probabilistic neural network (PNN) algorithm has been presented and we use the performance comparison between these 8 models to illustrate where financial-services customers fall into a category of good or bad based on the different techniques. Because of this combined technique, the banking system, especially the weak banking system, can categorize its customers into good and bad ones. In fact, the purpose of this paper is to create a hybrid approach for credit scoring Iranian banks' clients, thus obtaining the probability of default and credit risk models for the banking system.