The Application Of Selective Logistic Regression Ensemble Algorithm In P2P Net Loan Credit Evaluation

Ensemble learning is one of the hot topics in the field of machine learning in the past twenty years.The principle of ensemble learning is to improve the accuracy and stability of the model(the following are referred to generalization ability).The theoretical analysis shows that,for the given classification task,after the use of integrated learning more a base classifier,a better generalization performance is obtained by selecting a part from the base set of classifiers for integration than using all base classifiers in certain conditions.Therefore,selective ensemble learning has become an important research field.Logistic regression is one of the most commonly used methods of scholars,so this paper designs the logistic regression into the selective ensemble algorithm.On the one hand,this article through to improve the prediction effect of single model to improve the generalization ability of ensemble learning model;on the other hand,this paper selectively integrated part of the base learner results further improve the generalization ability of model.Through analysis of the basic user net loan data provided by big data platform,and compared with the single logistic regression model and GBDT,results show that 10-fold cross-validation of the Logistic regression model,GBDT algorithm and selective Logistic regression algorithm,the AUC average are 0.585,0.600 and 0.601.Respectively,the method presented in this paper is superior to other two algorithms.