Social Data Driven Risk Assessment Model In P2P Leading

P2P lending is a new internet banking form,and it has become an important choice of personal finance,but there is a default problem in P2P lending.To solve this problem,most existing recommendation model only focus on information of the bid and the bid’s owner.However,different borrowers are interrelated,so we introduce the concept of socialization and use the Gradient Boosting Decision Tree(GBDT),which is a better classifier in the enhanced learning model.Then we propose a social data driven risk assessment model in P2P leading(GBDT-SOC).First of all,in this paper,we through the data statistics,found the potential relationship between social factors and the borrower default.We establish the social influence model use lender’s peer-to-peer relationship and the association between groups based on group.In addition,we use feature engineering to reconstruct feature and filter feature,then use the GBDT to model these features,last combine our social influence model with GBDT to build our model.Finally,we train parameters with gradient descent method and assess the risk of borrowing.The experiments are based on the data from Prosper which is well-known P2P platform.Experimental results show that,compared with the traditional P2P loan risk assessment model,our model have a great performance over other model in give data-set.