A Prediction Model for Recognition of Bad Credit Customers in Saman Bank Using Neural Networks

The aim of this paper is to present a model based on feed forward neural networks to recognize bad credit customers in Saman Bank. To find an appropriate structure for the proposed neural network model, three different strategies called quick, dynamic and multiple strategies are investigated. The registered data of credit customer in Saman Bank from 2000 to 2008 year is used. To prevent models from over fitting with training data specifications, according to cross validation, we divide existing data set into three subsets called training, testing, and validation set, respectively. To evaluate the proposed model, we compare the result of three different strategies in neural networks with each other and with some common prediction methods such as decision tree and logistic regression. The results revealed that the threelayer neural network based on the back propagation learning algorithm with quick strategy has higher accuracy.