Mining the customer credit by using the neural network model with classification and regression tree approach

A combination of classification and regression trees (CART) and neural network techniques is proposed to determine whether the predictive capability can be enhanced in a credit-scoring model. To demonstrate the effectiveness of the proposed approach, these techniques are applied to data from a large bank in Taiwan. In the neural network and the combined model approaches, the backpropagation learning technique, with various learning rates, was extensively studied to determine the connection weights between the neurons. The number of hidden neurons was also varied to determine its effect on the convergence rate. Our results indicate that the proposed combined approach predicts much more accurately and converges much faster than that the conventional CART method or the neural network approach