Prediction of loan default based on multi-model fusion

With the development of Internet technology, online loans continue to enter the public eye, individuals and small businesses must access to more loan opportunities, and it is important for online loan platforms to effectively reduce the credit crisis associated with customer loan defaults. This paper uses the loan default dataset from lending club. The ADASYN (Adaptive synthetic sampling approach) method is adopted to cope with the class imbalance problem of the dataset. In order to improve the prediction accuracy, this paper utilizes the Blending method to fuse three models: Logistic Regression, Random Forest, and CatBoost. After experimental comparison, it is found that the performance of the fusion model proposed in this paper is better than the three models of Logistic Regression, Random Forest, and CatBoost, which can effectively predict the probability of customer loan default through the training of the dataset and reduce the external risk brought by the online loan platform facing customer loan default.