A rejection inference technique based on contrastive pessimistic likelihood estimation for P2P lending
作者：
Highlights：
• We propose a novel reject inference model for peer-to-peer learning.
• The proposed model uses rejected samples to model probability of default.
• The proposed model integrates semi-supervised learning with gradient boosting.
• The results implemented on two real-world datasets show the efficiency of the proposal.
• We reveal that the rejection rate affects the predictive accuracy of the proposed model.
摘要
•We propose a novel reject inference model for peer-to-peer learning.•The proposed model uses rejected samples to model probability of default.•The proposed model integrates semi-supervised learning with gradient boosting.•The results implemented on two real-world datasets show the efficiency of the proposal.•We reveal that the rejection rate affects the predictive accuracy of the proposed model.
论文关键词：Big data applications,Contrastive pessimistic likelihood,Credit scoring,Data analytics,Gradient boosting decision tree estimation,Machine learning,P2P lending,Reject inference,Semi-supervised learning
论文评审过程：Received 24 May 2018, Revised 25 May 2018, Accepted 25 May 2018, Available online 26 May 2018, Version of Record 7 June 2018.
论文官网地址：https://doi.org/10.1016/j.elerap.2018.05.011
