Design And Implementation Of Credit Risk Prediction System For Banks Based On Ensemble Learning

The credit business of a commercial bank has always been one of its core businesses,and due to the uncertainty of its own behavior,asset status,and external factors,the bank's service providers have too many uncertain factors and the credit risk has also increased.Nowadays,the credit management of SMEs and individual customers is generally operated by account managers in accordance with bank internal standards.They have absolute control over the credit management to their credit subjects,which exposes customers to information security,artificial credit,privacy protection,financial risk,etc.Credit risk prediction helps commercial banks to control capital risks,taking measures to reduce risks in the main time,reducing adverse consequences caused by them,and ensure the stable and healthy development of the banking industry.Therefore,research on credit defaults is of great significance to commercial banks.Aiming at the default of the personal credit business of commercial banks,this paper designs a commercial bank credit risk prediction system based on an integrated learning model,which depends on research of machine learning and other related technologies and the characteristics of the credit risk of commercial banks.Among them,the research focuses on credit risk prediction models,multi-angle feature derivation of customer credit data,and feature selection in terms of evolutionary algorithms for features with large dimensions and complex associations.According to three popular integrated learning algorithms,including Adaboost,Xgboost and Lightgbm,the fusion model is constructed by using voting method.On this foundation,the paper combines the Python and Django framework to deploy the optimized model to the credit risk prediction system.The system mainly consists of system management,risk prediction,risk disposal and other modules.Among them,the system management module contains user management,data management and other functions;the risk prediction module is used to assess the risk of customers,the risk disposal module is used for the customer manager to give the disposal suggestions,and set the risk rating.The purpose is to combine machine learning with the credit management system of ordinary commercial banks,and apply machine learning technology to the credit risk management,so as to assist bank account managers in assessing loan risk.The experimental results show that the accuracy of the prediction model based on ensemble learning is great,and the works will strengthen the reliability and efficiency of the credit risk management of commercial banks and reduce the human factors of credit management,which is beneficial to improving the level of bank credit business.