SVM Customer Credit Evaluation Model Based On Bayesian Network

Limited by data collection conditions,there is a serious lack of data in the credit evaluation of bank customers.It will have a great impact on subsequent credit assessment.The commonly used methods for missing values are to delete the records with missing data or to fill the missing data.Obviously,missing data filling will lay the foundation for future credit assessment.In this paper,a method of missing data filling based on Bayesian network and probabilistic inference is proposed.According to the existing data,Bayesian network is established by mining the correlation between attributes,and then it will simplify solution of multidimensional joint probability.According to the results of probabilistic reasoning,the missing data is filled,The results also reflect the accuracy of data filling.The higher the probability,the higher the accuracy of data filling.Using the processed data,the customer credit evaluation system based on support vector machine is established ? The credit evaluation model is three stages,including decision model of default,calculation model of default probability and prediction model of customer credit risk.The customer credit can be evaluated more comprehensively.In order to reduce the bank's bad credit rate,It has important theoretical and practical significance.