Credit Risk Assessment Based On Machine Learning

Credit business is one of the most important parts of a bank's many businesses,and it has a very important impact on the bank's earnings and development.The most important part of the credit business is the assessment of credit risk.Accurate assessment can enable banks to increase their profits with the lowest possible risk.With the rapid development of information technology in the new era,banks have ushered in the era of big data.In this era,banks have more and more customer information.In order to effectively use these customer information data,the bank will analyze the corresponding business based on the customer information data to help decision-making.This paper selects the German credit data set,establishes the four models of multiple Logistics regression model,discriminant analysis,naive Bayes and neural network,and then analyzes the prediction results of the four models separately.In addition,the accuracy rate,Recall rate and precision rate and other evaluation indicators,evaluate and compare the evaluation results of the four models.Through the comparison of the various indicators of each model,the following conclusions are finally drawn: for the analysis of credit risk assessment problems,the established multi-Logistics regression model,discriminant analysis,naive Bayes and neural network four models,evaluate the effect.The best model is the neural network model.Therefore,in the future analysis of credit risk problems,the analysis can be based on the neural network model,so as to establish or improve a new model that is more conducive to the analysis of the problem.