Effective Matching for P2P Lending by Mining Strong Association Rules

Abstract

Disrupting traditional financial models and businesses, FinTech integrates both finance and technology, provides an array of innovative business services, and leads the revolution of global economy. Nowadays, main business models in FinTech are third-party payment, peer-to-peer (P2P) lending, and crowd-funding. P2P lending is the work of lending money to individuals or small and medium-sized enterprises through online services that match lenders and borrowers directly within websites. The matching considers risks and requirements of both lenders and borrowers, offers lenders with attractive return rates and credit worthy borrowers, and provides the service more cheaply than traditional financial institutions with lower overhead and threshold. Hence, P2P lending is the key and an important trend of Fintech. However, without financial institutions, P2P lending will cause risk management problems including credit risk, business risk, and market risk. Unfortunately, no adequate regulation are provided to protect the unsecure personal loan due to the rapid progress and beyond the laws. Still, P2P lending platforms enable borrowers to propose expected interest rates and lenders to reduce transaction risks, and greatly improve matching efficiency. Therefore, this study mines the association rules from the famous P2P lending website Zopa by analyzing basic member data and past transactions. The discovered associations and distributions can further be used in suggesting the optimal decisions for the borrowers so that the matching will be more effective. In this paper, the borrowers' data in the P2P platform is targeted, factors including the total number of payments, interest collected, terms, lending rate, latest status, and postcode are extracted to assist matching the suitable borrowers so that both parties may have higher transaction satisfaction.