A Bayesian Investment Model for Online P2P Lending - Frontiers in Internet Technologies - page 21

A Bayesian Investment Model for Online P2P Lending *
Xubo Wang, Defu Zhang, Xiangxiang Zeng ** , and Xiaoying Wu
Department of Computer Science, Xiamen University, Xiamen, China
[e-mail],{dfzhang,xzeng}@xmu.edu.cn,
[e-mail]
Abstract. P2P online lending is an emerging economic lending model. In this
marketplace, borrowers submit requests for loans, and lenders make bids on
them. It has put forward new challenges to investors about how to make effec-
tive investment decisions. Bayesian network is a probabilistic graphical model
that represents a set of random variables and their conditional dependencies. In
the paper, we calculate the mutual information of every two variables to meas-
ure their mutual dependence and build a Bayesian network model to select
loans that would pay back with high confidence. We perform abundant experi-
ments on the data from the world's largest P2P lending platform Prosper.com.
Experimental results reveal that Bayesian network model can significantly help
investors make better investment decisions than other investment models.
Keywords: P2P lending, Classification, Bayesian network, Tree Augmented
Naïve Bayesian.
1 Introduction
P2P lending , also called online social lending, allows direct lending and borrowing
between individuals on an Internet-based platform, without the participation of
traditional financial intermediaries such as banks (Wang, 2009). In this way, it pro-
vides convenient online services for reallocating small funds in credit transaction.
There are more than 30 online P2P lending markets in more than 10 countries in the
world, such as Zopa in UK and Prosper in the US. In recent years, advances in P2P
lending marketplaces have provided new research opportunities with the availability
of massive amounts of P2P transaction data. In this study, we focus on Prosper
(http://www.prosper.com), the largest online P2P lending market in US, which has
helped its 1.26 million members receive over $314 million loans. In this marketplace,
borrowers submit requests for loans (called listing), and then lenders make bids on
them. Prosper handles the aggregation and disbursement of funds to borrowers and
then services the loans, collecting and distributing payments and interest back to the
loan investors.