Research On P2P Network Loan Investment Decision Based On Improved Bipartite Graph

P2P network lending is a new type of lending mode in the era of Internet finance.It provides more convenient and efficient financing channels for individuals and small and micro enterprises in a unique online way.In recent years,P2 P network lending has developed rapidly,but with the implementation of national policies and the strengthening of supervision by regulatory authorities,there are more and more P2 P network lending platforms going out of business,running away and transforming.For the investors involved,due to the virtual nature and information asymmetry of network lending,this will increase the investment risk faced by investors.Due to the complex relationship between the investors and borrowers of P2 P network lending platform,the complex network can be introduced into P2 P lending research,which provides a new research idea for P2 P network lending.Therefore,in view of the characteristics of many to many loan relationship in network lending,it is of great theoretical and practical significance to use machine learning method and decision-making model to study how to reduce investment risk and obtain more profits for investors.In this paper,we first recommend a suitable list of borrowers to investors in a progressive way,and then provide investors with a fund allocation scheme on the list.Specifically,first,through the P2 P network lending platform investors and borrowers' many to many investment relationship,considering the investment amount and investment time,an improved bipartite graph method is proposed to recommend suitable borrowers for investors.Then the logistic regression model is constructed to predict the default probability of the borrower.On this basis,a mean variance model of risk preference coefficient is proposed to build the P2 P lending investment decision-making model,which provides the investors with the investment decision-making scheme.At last,the two-stage decision-making model proposed in this paper is validated by the experimental study of P2 P lending platform data in China.The results show that the bipartite graph recommendation algorithm with investment amount and investment time can effectively improve the recommendation accuracy and recall rate,and then the mean variance model with risk preference coefficient can help investors get better returns.This paper proposes a two-stage method of recommending suitable borrowers first and then combining them to provide fund allocation scheme,which has achieved good practical results,and has certain reference significance for the research of investor risk and income in P2 P network lending scenario.