BP Neural Network And Logistic Regression In The Application Of Credit Rating Compared With The Model

The financial industry,with the rapid development of economy,technology,banking business,the increasing maturity of the combination of science and technology and traditional industry has become the mainstream of The Times.Also with p2 p,the bank on the net,raise the emergence of lending platforms,such as the bank personal credit evaluation in the financial industry development status is also more are important.For the healthy,and quickly development of economy,credit plays the significant role in it.the bank of the personal credit assessment also affects the housing loan,car loan and so on into the business,so the fast speed,high efficiency of credit evaluation have become the common demand of most Banks improve work efficiency.In an age of rapid development of science and technology,the use of artificial intelligence search platform for more information on credit evaluation,as well as in the driving of the maximization of profit and risk minimization,the use of data mining,and machine learning algorithms,can significantly improve the efficiency of the examination and approval,colleagues to avoid lending bank customer churn and negative influence of the error.This article from the development and current situation of the bank credit evaluation analysis,the article 1000 in the UCI credit data as data sources,using data mining,and machine learning algorithm,model is established.Using machine learning algorithm of neural network,and Logistic regression,among them,70% random sampling to select data as the training set,while the rest of the data for the test set,used to trained model prediction,prediction results and compared with the real value,calculation model prediction accuracy,and model evaluation.On the model of evaluation,by using the testing method is mainly the ROC curve,LIFT charts,K-S curve and index of AUG to compare the effect of two kinds of model of strength and stability,and gives the corresponding parameter estimation,statistical figure,etc.,and the final of the comparison results and the practical significance of the model.This article will the machine learning algorithms of data mining combined with actual data bank,the choice of the bank’s credit evaluation method has certain reference value.