Personalized Recommendation in P2P Lending Based on Risk-Return Management: A Multi-Objective Perspective

P2P lending is an increasingly prosperous financial market, where lenders can directly bid and invest on the loans posted by borrowers. However, when facing massive loan requests, it is very difficult and also boring for lenders to choose loan portfolios meeting their ideal expectations. Actually, when choosing loans, most lenders pursue the highest profit with the lowest risk as well as satisfying their hobbies. In this paper, we formalize a multi-objective optimization problem to help lenders select loan portfolios. Specifically, the recommending scenario in P2P lending is formulated as a multi-objective optimization problem, where two objective functions are designed for capturing lenders' multiple demands. On this basis, a multi-objective evolutionary algorithm based on return-risk management named MOEA-RRM is then proposed for the multi-objective optimization problem, which can help lenders choose loan portfolios to meet lenders' multiple demands. Furthermore, in MOEA-RRM, a decision space dimensionality reduction strategy and an initialization strategy are proposed to improve the performance of algorithm. Finally, experimental results on a real-world P2P lending data set demonstrate the effectiveness of our proposed MOEA-RRM, i.e., the proposed approach can recommend the loan portfolio with a good trade-off between risk and return as well as satisfying the hobbies of lenders.