Research On Personalized Investment Recommendation In P2P Lending

For potential investors in P2 P lending market,there are two key issues which need to be solved.One is how to choose the right investment projects considering their investment needs and preferences;The other is How to reasonably allocate their investment amount to these projects.However,decision-making information that investors are faced is soaring with growing scale of platform development and increasing variety of loan projects.Because of ―information overload‖,incompleteness and asymmetry of information related to key decisions and market noise,investors,especially many non-professional investors,often face serious adverse choices and then fall into decision-making dilemma.Personalized investment recommendation can save the cost of investors’ information search,improve the ability and efficiency of investors’ information identification,inquiry and analysis,help investors get rid of the dilemma of decision-making,and assist investors to make effective decisions.At present,there are few researches on personalized investment recommendation of P2 P lending.which mainly focus on default risk prediction,investment similarity calculation and investment capacity analysis of projects and so on,and often neglect to fully explore the knowledge closely related to investors’ investment decision-making behavior from the perspective of investors’ actual decision-making behavior characteristics and decision-making psychology.As a result,it is difficult to accurately grasp the investment trend of investors and improve the effectiveness of investment recommendation.In view of this,in order to improve the performance of investment recommendation in P2 P lending market,promote the long-term sound development of the market,this thesis analyzes the factor influencing investors decision-making psychology and decision-making behavior according to the characteristics of market investors’ behaviors and key decision-making issues,excavate the key knowledge which is helpful to grasp investment demand and decision preference of investors from the perspectives of social network,herd behavior and investor’s risk preference based on the platform historical transaction data and the relationship between different objects,that is,key influencing factor of investors’ decision-making behavior,and apply them to the design of investment recommendations.Main research contents include:(1)Building object association network model of P2 P lending market and then excavating risk characteristics of friend network based on friend relationship of borrowers.Due to the incomplete and asymmetric information of P2 P lending market,when investors make investment decisions,a large amount of market information and economic signals transmitted by these correlations will have an important impact on their investment strategy choices.based on historical transaction data and related object property information in Prosper platform,various associations among key objects are analyzed,and the object association network model of P2 P lending market is constructed;Secondly,based on object association network model of P2 P lending market,the candidate features of friend network related to the project are comprehensively explored,and the friend network features,which are significantly related to project default status,are screened out and introduced into the P2 P lending project default risk prediction model to provide support for optimizing portfolio recommendation effect.(2)Designing the influence factor of friend bidding behavior based on investor friend relationship and then establishing a personalized investment recommendation framework in P2 P lending considering friend relationship.Investment decision-making behavior of investors is not only related to the investment similarity of projects and investment ability of project investors,but also influenced by the investment behavior of other market participants(especially direct friends with high investment ability)in online social networks to a large extent.First,considering the influence of the bidding behavior of investors’ direct friends on their investment preference and decision-making behavior,we analyze friend relationship and investment behavior relationship among investors based on the object association network model of P2 P lending market,explore the impact of bidding behavior of investors’ direct friends on their investment preferences,and design the influence factor of friend bidding behavior;Secondly,this factor is applied to the prediction of investor’s investment interest degree to improve the effectiveness of investment project recommendation;Thirdly,in order to improve the effect and economic performance of investment recommendation,an investment recommendation scheme considering friend relationship in P2 P lending market is designed based on risk characteristics of friend network and friend relationship among investors.(3)Constructing the influencing factor of rational investment behavior and then designing a personalized investment project recommendation scheme from the perspective of market herd behavior and rational investment.For most investors in P2 P lending market,especially non-professional investors,when obtaining and effectively analyzing relevant decision information(such as project default risk prediction information)requires a high cost,in order to reduce expected investment risk due to adverse selection,many investors often seek low-cost market signals(relevant information which can reflect strategy choice of most investors in the market)as a decision basis,sometimes even completely ignore their own private information to herd.In the P2 P lending market,investors tend to be influenced by the market herding behavior to some extent when making investment decisions.Based on traditional collaborative filtering recommendation method,the influence of herd behavior on investors’ decision-making behavior is further considered.On the one hand,considering the general herd mentality of investors in P2 P lending market,the herd behavior tendency of each investor which can reflect his/her strength of herd mentality is calculated based on object association network model of P2 P lending market.on the other hand,because indexes of project herd degree in previous studies which contained a lot of market noise are unable to gather popular wisdom,guided by rational investment in the market,we excavate characteristics associated with project herding degree based on object associated network model of P2 P lending market and bidding correlation data between investors,and select the characteristics that are significantly related to default risk of projects,namely risk characteristics of project herding degree,and then design rational investment behavior influence factor.By comprehensively considering investment similarity,herd mentality of investors and rational investment driving force of projects,the recommendation scheme of personalized investment projects in P2 P lending is designed to improve recommendation effect and further guide rational investment of investors.(4)Constructing a portfolio optimization model considering expected utility maximization of investors.Theory and practice of portfolio investment show that investors are cautious and conservative to risk in uncertain market environment.that is,most of them are risk averter,and have different risk aversion degrees.In the financial field,when faced with uncertain environmental conditions,investors with different risk aversion degree often tend to choose the decision plan that maximizes their expected utility.Considering the influence of risk preference of investors in P2 P lending market on their decision-making behavior,risk aversion coefficient and risk tolerance of investors are investigated on the basis of Prosper platform’s historical transaction data.Based on expected utility theory,a portfolio recommendation algorithm based on project risk management and expected utility maximization of investors is designed.The innovative work of this thesis is concluded as follows:(1)Constructing object association network model in P2 P lending,exploring the conceptual features related to investment decisions and then defining the influencing factors of investors’ decision-making behavior.According to the behavior characteristics of investors and key influencing factors of decision-making,we analyze various correlation relationships among related objects based on historical transaction data of P2 P lending platforms and attribute information of objects,and then build association network model of objects in P2 P lending market.On this basis,the key conceptual characteristics of investors and projects related to investment decisions are explored from the perspectives of social capital,herd behavior and risk preference,and corresponding conceptual models are constructed,which will provide support for the discovery of knowledge related to investment decision.Firstly,excavating risk characteristics of friend network based on friend relationship of borrowers to improve the validity of project default risk prediction and provide support for the subsequent portfolio optimization.Secondly,considering the influence of the bidding behavior of investors’ direct friends on their decision-making behavior,the influence factor of investor’s bidding behavior based on investor’s friend relationship is designed to effectively grasp investors’ interest and preference in investment decision-making.Thirdly,considering the influence of herding behavior on decision-making behavior of investors,the influencing factor of rational investment behavior is designed to improve the quality of investment project recommendation,and guide investors to make rational investment.Finally,Considering the influence of investors’ risk aversion degree on their decision-making behavior,the risk aversion coefficient of investors is calculated to provide support for the subsequent portfolio optimization.(2)Based on the key influencing factors of investors’ decision-making behavior,two personalized investment project recommendation methods are designed.On the basis of the traditional collaborative filtering recommendation method,we introduce the corresponding key influencing factors of investors’ decision behavior for their investment project selection,and respectively construct recommendation scheme of personalized investment projects in P2 P lending market from the perspective of social capital and market herd behavior.They are the investment project recommendation method considering the friendship between investors in P2 P lending market and the investment project recommendation method considering herd behavior and rational investment in P2 P lending market.Two new evaluation indexes are proposed to comprehensively evaluate investment recommendation quality of this method.(3)Based on the key influencing factors of investors’ decision-making behavior,two portfolio optimization methods are designed.On the basis of the traditional P2 P portfolio recommendation algorithm,we further consider the influence of differences in investors’ risk preference and assessment effects of project default risk on allocation of investors’ capital and introduce the corresponding key influencing factors of investors’ decision-making behavior.Aiming at allocation of investors’ capital,portfolio recommendation models in P2 P lending are respectively constructed from the perspective of social capital and expected utility maximization.They are the portfolio recommendation method considering friend relationship of borrowers and the portfolio recommendation method considering expected utility maximization of investors.Two new evaluation indexes are proposed to evaluate investment decision satisfaction,economic benefit and utility of the model.The recommendation results of our models are compared with other benchmark models to comprehensively evaluate their recommend effect.We implement experiments on real dataSets of Prosper platform,experimental results demonstrate that our methods have better recommendation quality than traditional investment recommendation method.