A Study on P2P Lending Using Data Mining Technology

• Main Authors: Kuo,Chia-Hsin, 郭佳馨
• Other Authors: Hsueh,Sue-Chen
• Format: Others
• Language: zh-TW
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/v8rk8t

碩士 === 朝陽科技大學 === 資訊管理系 === 106 === Innovations and rapid developments of information technology have changed traditional consumption models. Particularly FinTech (Financial Technology), which integrates both finance and technology to provide an array of innovative business services, is revolutionizing the global economy. FinTech is characterized with several service models, such as Third-Party Payment, Crowdfunding, Robo-advisor, and P2P lending. Among others, P2P lending is a microcredit activity of loans or investments without engagements of financial institutions on Internet. Through diversified credit mechanisms and risk managements, the transaction needs of both borrowers and investors can be fulfilled. The participants of P2P lending usually are individuals or small and medium-sized enterprises, the huge number of participants and the low thresholds of monetary amounts enable fast and direct matching of both parties. However, having no financial intermediaries and low thresholds also bring about potential credit and business risks. Therefore, this study aims to provide secure mechanisms of P2P lending to reduce the risks. The progress of P2P lending is so fast that no adequate regulations or laws may catch up to protect the unsecured personal loan, which situation puts both investors and loaners at high risks. In addition to profits, both parties have to consider issues of time and costs, risk of taking bad debts, and fail recovery during mediation. This may result in a decrease in efficiency and effectiveness or a failure to meet the demand of borrowing and investment. P2P lending cannot bring a real benefit into full play without solving these issues. In this study, we have analyzed the profiles and transactions of P2P lending members from the P2P lending ancestor Zopa and the largest club Lending Club. The group datasets with large amounts of data or parameters are preprocessed by a K-means clustering algorithm to segment various types of members. We then applied the Apriori algorithm to find out the correlation and distribution of data between transactions. Thus, P2P-lending users may quickly grasp the relationships between transactional data. The discovered rules enable both borrowers and investors to perform matchmaking more effectively. Meanwhile, suitable investment decisions can be provided for investors. In addition, platform operators may grasp the operation of P2P lending platforms more accurately.