With the development of Peer-to-peer credit industry in recent years, more and more consumers choose to solve the problem of temporary shortage of funds through Peer-to-peer loan. Unfortunately, although this is conducive to the allocation of funds, but the loan applicants for various reasons overdue by the phenomenon of many times, this gives Peer-to- peer company a certain degree of loss of bad debts. The machine learning algorithm can be used to discriminate the applicants, to deny the users with great overdue risk before the loan is issued, it could control the financial risk and reduce the risk of the enterprise's potential bad debts effectively. Because I am now working for a P2P company,and i hope that my studies can be better serve my work, so i choose this project as my graduation project.In this project, i will use the well- known Peer-to-peer platform LendingClub Open loan Applicant data for the pre-loan applicants overdue classification model.