Design And Implementation Of Personal Credit Decision Model Based On Integrated Learning

With the continuous improvement of China's market economic system,more and more individuals are gradually accepting the consumption model of personal credit,and the share of personal credit in bank loan business is also rising.However,due to the lack of personal credit,the risk of personal credit has gradually become prominent.Personal credit default has caused a serious impact on the bank credit business and has become a hindrance to the rapid development of banks.How to effectively predict whether a borrower can repay on time is of great significance to commercial banks.This paper analyzes the needs of personal credit business and combines existing system development tools to design and implement a personal credit decision system based on integrated learning algorithms.The personal credit decision system designed and implemented in this paper includes various parts such as personal credit evaluation,database design and system management interface.The personal credit evaluation module adopts the current mainstream integrated learning algorithm,and uses the four models of support vector machine,random forest,XGBoost and LightGBM as the base learner,and the model is merged by means of relative majority voting to obtain the personal credit decision model.At the same time,the customer credit index extraction,data preprocessing,standardization and other data processing operations are completed.Finally,the object-based design method is used to encapsulate the personal credit decision model,and a unified calling interface is provided.The database design mainly includes the employee information table,the customer information table,the model prediction result storage table,and the design of the model hyperparameter tuning table.The system operation interface uses Java Swing technology based on C/S architecture to provide an interface for the use and management of personal credit decision system,including system user management and authority verification,customer data management and update,and evaluation result query.Finally,the personal credit decision system designed and implemented is tested and verified in the simulation environment.The results show that the designed system can achieve the expected results,and the system has high prediction accuracy,which can provide reference for manual review.