Marjan Rahbarfarazi - Department of Information Technology, faculty of computer, University of Applied Sciences & Technology(UAST), Tehran, Iran
Mohamad Reza Azadkhah - Department of Computer Engineering, Faculty of Engineering, University of Shahed , Tehran, Iran

Nowadays, evaluating the credit of customers is very important due to the increase in financial transactions of individuals with banks and financial institutions. Today, with the advancement of technology and population growth, it is practically impossible to use traditional methods to identify reliable customers. Therefore, it is necessary to provide a method that can judge the creditworthiness of customers using the information in their financial records and biographies. Therefore, various algorithms have been presented by researchers to model the credit of customers. In this article, an approach based on the combination of data balance algorithms and classification algorithms is presented to recognize the credit of bank customers. In this method, random subsampling approach is used to solve the problem of data imbalance in the model training process. Each of the extreme learning machine method models are trained on a balanced subset of data selected in this way, and finally, their prediction results are combined in a windy way. The final goal in this research is to determine the creditworthiness of bank customers. Two standard and related datasets (German Credit Data and Australian Credit Approval) from the UCI dataset have been used to evaluate the proposed method. The performance accuracy results of the proposed method in the German Credit Data set show an accuracy of ۶۸.۰۲ and an average readout of ۵۳.۳۱, and for the Australian Credit Approval data set, an average accuracy of ۸۹.۷۴ and an average readout of ۷۳.۹۴

validation - bank customers - data mining - data balance - ranking algorithm