A BP Neural Network Approach on Risk Evaluation of Ventures in China Customs

With the application of China´s customs management information system (MIS), a great deal of data that reflects the activities of import and export ventures has been stored in the database. However, the risk management of ventures in China customs still depends on the personal analysis of customs staff, which lacks the necessary decision support information. On the basis of the data warehouse of risk management system, a BP neural network model, combined with Levengberg-Marquardt algorithm, is adopted and implemented by Matlab 7.0 for the ventures risk evaluation in China Customs. The simulation analysis of the evaluation model demonstrates that the algorithm can work effectively to perfect the quality of risk controlled by the customs.