Application Of Data Mining Algorithm In Credit Score Model

The rapid development of the economy and society stimulates tremendous de-velopment of credit consumption,and various consumer loans have rapidly increased,and domestic commercial banks have expanded their businesses in credit consump-tion.Due to historical reasons,most commercial banks lack an effective method of personal credit scoring,and even if they have the disadvantage of poor evaluation accuracy,they have an adverse effect on the development of personal credit consump-tion.In addition,for commercial banks,because of the homogeneity of products and services,only by grasping the rules behind information from a large amount of busi-ness data and making rational decisions can we win in the fierce market competition.Traditional credit scoring systems,such as Naive Bayesian algorithms,are often based on a series of more rigorous assumptions,such as the independence of their respective variables,and it is difficult to achieve satisfactory results in the face of massive amounts of data with multiple collinearity.With the rapid development of computer and technology and data mining theory,the method of credit scoring model has been greatly enriched.Data mining technology has been widely used in the financial field in recent years due to its good ability to fit high-dimensional data.The research goal of this paper is to establish a personal credit scoring model based on the most widely used classification algorithm in data mining technology-decision tree classification model,and compare it with the results of Logistic re-gression modeling.After completing a series of data preparation tasks such as data missing value processing,sample extraction,and variable selection based on chi-square test and mutual information,the sample size of 3:1 is set as the training data set and the test data set,respectively.The methods were modeled separately and the accuracy,ROC,AUC and other indicators were used to evaluate the model.The conclusion that the credit score model obtained by using the decision tree algorithm was superior to the Logistic regression model.