The Scoring Matrix Generation Method and Recommendation Algorithm in P2P Lending

Abstract

With the rapid development of P2P lending in recent years, how to recommend bidding items to the bidder to meet its preferences will become a research hotspot, but there are few studies in this area. And since there is not a dominant scoring mechanism in P2P lending platform, it is an urgent to solve the problem of how to generate a reasonable score matrix. First of all, Crawling over more than 320 thousand tenders of the last two years from the PPDai.com to. Then, based on the deeply analysis of the data characteristics of the P2P lending areas, three different types of scoring matrix (Boolean, continuous and hybrid) are generated. Then, three kinds of classical collaborative filtering recommendation algorithms are used in the experiment. The results show that different types of scoring matrix has important influence on the performance of the recommendation algorithm, and the performance of cooperative filtering algorithms is different when applying in P2P lending compared with applying in some other fields. Finally, the experimental results are analyzed, and the relevant suggestions are given for the P2P lending.