A Deep Learning Approach for Credit Scoring of Peer-to-Peer Lending Using Attention Mechanism LSTM

With the rapid development of the peer-to-peer lending industry in China, it has been a crucial task to evaluate the default risk of each loan. Motivated by the research in natural language processing, we make use of the online operation behavior data of borrowers and propose a consumer credit scori...

Full description

Saved in:
Bibliographic Details
Published inIEEE access Vol. 7; pp. 2161 - 2168
Main Authors Wang, Chongren, Han, Dongmei, Liu, Qigang, Luo, Suyuan
Format Journal Article
LanguageEnglish
Published Piscataway IEEE 2019
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Subjects
Algorithms
attention mechanism
Credit scoring
Deep learning
Default
Feature extraction
Industries
LSTM
Machine learning
Mathematical model
Natural language processing
Neural networks
P2P lending
Peer to peer lending
Predictive models
Online AccessGet full text

Cover

More Information
Summary:With the rapid development of the peer-to-peer lending industry in China, it has been a crucial task to evaluate the default risk of each loan. Motivated by the research in natural language processing, we make use of the online operation behavior data of borrowers and propose a consumer credit scoring method based on attention mechanism LSTM, which is a novel application of deep learning algorithm. Inspired by the idea of Word2vec, we treat each type of event as a word, construct the Event2vec model to convert each type of event transformation into a vector and, then, use an attention mechanism LSTM network to predict the probability of user default. The method is evaluated on the real dataset, and the results show that the proposed solution can effectively increase the predictive accuracy compared with the traditional artificial feature extraction method and the standard LSTM model.
ISSN:2169-3536
2169-3536
DOI:10.1109/ACCESS.2018.2887138