A Comparison of Data Mining Approaches on Predicting the Repayment Behavior in P2P Lending

摘要

Default loan detection for P2P lending market is of significant importance with the present-day expansion of the market share of P2P loans. Besides general data mining methods for default loan detection, novel data mining techniques, such as XGBoost, AdaBoost and Random Forest approaches have gained increasing attention and have achieved better performances on the distinction between default loans and normal loans. Due to the development of data mining, most of these new techniques have attained more accurate results than the previous data mining techniques. However, very few studies discuss the performance of difference data mining approaches in a thorough manner, from data cleaning to the final results, for predicting the loan statuses. Thus, in this paper, we provide a thorough data analysis process for predicting the loan statuses by investigating the performance of the above mentioned approaches.