Analyze On The Factors Of Successful Borrowing And Investment Decision Recommendations On P2P Network Platform

In recent years, the rapid development o f P2P network lending, has become an important choice for investors to allocate assets. P2P network platform was born, which is a direct transaction between investors and borrowers. Compared to traditional financing and investment channels. P2P network pla tform is more convenient for investors to bring higher income. But facing a number of loans, investors have been able to determine whether the borrowing is successful through information.At the same time investors can effectively evaluate the benefits and risks of the loan, so that to obtain the desired income, which is more worthy of our attention. C urrently the research on P2P network lending mainly concentrated in lending practices and supervision level, lacking of standing in the investor point determine the successful borrowing rate and provide decision- making advice.This paper stands in the perspective of investors, trying to find an effective way for investors to measure the probability of success of the loan and investment efficiency. Therefore, firstly, based on the peer-to-peer platform of historical data, this paper construct the model of the influencing factors of the success rate of lending by a logistic regression model. Secondly, we use mathematical kernel weight model, according to the similarity metric, for new loans and history of borrowing similarity weights, new borrowing targets of return and risk, finally the application of Markowitz portfolio theory, calculated the loan portfolio for investors with different risk preference.The empirical results show that in the factors that affect the success rate of borrowing, investors are more concerned about loan amount, loan interest rates, loan term, the number of successful loan and flow loan. The ROC curve measurement have a good prediction. Therefore, investors can have an intuitive judgment by the borrow information, to determine whether it can be successful. Secondly in the kernel weight model through continuous testing to select optimal bandwidth, and we compared the linear regression of yield prediction model with kernel weight model, and the measurement of portfolio based on traditional credit rating model and clustering classification model. The results show that the kernel weight model has good applicability. Finally, according to the d ifferent risk preferences of investors, given the utility maximization investment scheme, the risk- free interest rate is introduced, so that to give investors more diversified choice.