Research of Credit Sale Risk Evaluation Based on BP Neural Network

Based on analyzing the current way of evaluating the credit risk, BP neural network (BPNN) is used in the paper. BPNN is powerful for the problem with non-linear and high dimension. A multi-class BPNN model is constructed to measure the credit sales risk. In this model, dominant factors affecting the credit sales risk are the input vectors and the credit risk is divided into five grades. Through a special data dealing process, the multi-class BPNN, trained by samples, identifies out the five grades of credit sale risk. An empirical analysis shows the proposed model is feasible and accurate.