Design And Implementation Of Bank Credit Risk Analysis System Based On Machine Learning Technology

Traditional commercial banks have accumulated a large number of user resources and user information in the long-term operation process.If these resources and information can be effectively used,commercial banks can produce a greater boost to product development.However,there are some technical barriers in the analysis and processing of customer information and the utilization of large data technology in commercial banks.In view of this,this thesis studies the application of data mining in credit risk analysis of commercial banks.1,The risk prediction model for bank credit has been constructed based on related theoretical research on machine learning by introducing a discovery method of association rule,in which cluster analysis is basically adopted to analyze client information and consecutive values are transferred to discrete variables of state property.The model will provide a foundation for association rule discovery afterwards as well as reference to clients' behaviors and habits for commercial banks.The commercial bank client information will be discovered by association rule based on Apriori algorithm,which will be used to explore the relations among clients' basic information,bank transactions and agreement violation.Based on the discovery of association rule,specific analysis subjects will be matched with former item in associate rule that is already explored,then clients' confidence in agreement violation in corresponding type of credit can be acquired to predict potential risks in credit loans for clients.2,The overall design and functional design of the system are carried out based on the risk analysis and prediction model.The system is divided into four functional modules: business information management,basic information management,credit information management and credit risk management.The flow charts and timing charts of different functional modules are given.3,Based on J2 EE technology,the scheme is implemented.The data samples of Bank Of China are used for experimental analysis.Using the accuracy rate as the basis of analysis,the risk analysis method proposed in this thesis is checked.The results show that higher prediction accuracy can be maintained at different levels of default confidence,and higher prediction accuracy can be maintained under different types and different amount settings.Commercial banks can achieve different intensity of risk control by taking different default confidence as the basis of credit granting.