基于图卷积神经网络的个人信用风险预测

doi:10.3778/j.issn.1002-8331.2211-0262

摘要/Abstract

摘要： 信用风险的评估与管理是金融机构的重要任务之一。为探究个人征信样本与违约样本间存在相似性时，能否体现个人征信的信用风险，并对预测其违约做出贡献。基于LendingClub 2020年第1~3季度贷款数据，利用GAMI-net从高维征信数据中筛选样本违约特征，并通过曼哈顿距离构建样本相似性网络，以反映样本整体间的征信相似性，并建立基于图卷积神经网络的信用风险预测模型。研究发现，个人征信相似性能体现信用风险，并对预测产生显著正面贡献。基于图卷积神经网络的风险预测模型在实验中的AUC值为81.60%，准确率为73.71%，相较于所有基准对比模型均大幅提升，表明考虑了样本相似性网络的图神经网络模型在信用风险预测精度上远优于未考虑样本相似性的机器学习模型。此外，所提供的特征筛选及样本间网络构建方法也为大数据智能风控提供了一定的借鉴意义。

关键词: 个人信用风险, 图卷积神经网络, 相似性网络

Abstract: Credit risk assessment and management is one of the important tasks of financial institutions. In order to explore whether the personal credit investigation data and default data are similar, can reflect the credit risk and make contributions to the prediction of the default, it takes the loan data of LendingClub from the first quarter to the third in 2020 as an example. By using GAMI-net to select customers’ default features from high-dimensional data, and constructing a loan customers similarity network through Manhattan distance to reflect the complex connection between customers, a personal loan default prediction model based on graph convolutional neural network is constructed. The AUC value of the prediction model based on the graph convolutional neural network in the experiment is 81.60%, and the accuracy rate is 73.31%, which is greatly improved compared with all the benchmark models. The graph neural network model considering the customer network is more superior to the machine learning models which does not consider the customer network in terms of loan default prediction accuracy. In addition, the methods of feature selection and network construction provided in the paper also offer a certain reference for the intelligent risk control methods in big data era.

Key words: credit risk prediction, graph convolutional neural network, credit similarity network

梁龙跃, 王浩竹. 基于图卷积神经网络的个人信用风险预测[J]. 计算机工程与应用, 2023, 59(17): 275-285.

LIANG Longyue, WANG Haozhu. Personal Credit Risk Prediction Based on Graph Convolutional Network[J]. Computer Engineering and Applications, 2023, 59(17): 275-285.

参考文献

[1] 王正位，周从意，廖理，等.消费行为在个人信用风险识别中的信息含量研究[J].经济研究，2020，55：149-163.
WANG Z W，ZHOU C Y，LIAO L，et al.Informational content of consumption behavior in consumer credit risk evaluation[J].Econimocial Research，2020，55：149-163.
[2] 曹再辉，余东先，施进发，等.两层分类器模型应用于个人信用评估[J].控制工程，2019，26（12）：2231-2234.
CAO Z H，YU D X，SHI J F，et al.The two-layer classifier model and its application to personal credit assessment[J].Control Engineering of China，2019，26（12）：2231-2234.
[3] 张润驰，杜亚斌，薛立国，等.基于相似样本归并的大样本混合信用评估模型[J].管理科学学报，2018，21（7）：77-90.
ZHANG R C，DU Y B，XUE L G，et al.A hybird large sample credit evaluation model based on combing similar samples[J].Journal of Management Sciences in China，2018，21（7）：77-90.
[4] 张娟，张贝贝.基于Group-LASSO方法的广义半参数可加信用评分模型应用研究[J].数理统计与管理，2016，35（3）：517-524.
ZHANG J，ZHANG B B.The application of generalized semi-parametirc addictive credit score model based on group-LASSO method[J].Journal of Applied Statistics and Management，2016，35（3）：517-524.
[5] 马晓君，沙靖岚，牛雪琪.基于LightGBM算法的P2P项目信用评级模型的设计及应用[J].数量经济技术经济研究，2018，35（5）：144-160.
MA X J，SHA J L，NIU X Q.An empirial study on the credit rating of P2P projects based on LightGBM algorithm[J].The Journal of Quantitative & Technical Economics，2018，35（5）：144-160.
[6] 刘志惠，黄志刚，谢合亮.大数据风控有效吗?——基于统计评分卡与机器学习模型的对比分析[J].统计与信息论坛，2019，34（9）：18-26.
LIU Z H，HUANG Z G，XIE H L.Is risk management with big data effective?—Comprison and analysis based on statistics score card and machine learning model[J].Journal of Statistics and Information，2019，34（9）：18-26.
[7] LUO C C，WU D S，WU D X.A deep learning approach for credit scoring using credit default swaps[J].Engineering Applications of Artificial Intelligence，2017，65：465-470.
[8] KIM J Y，CHO S B.Predicting repayment of borrows in peer‐to‐peer social lending with deep dense convolutional network[J].Expert Systems，2019，36（4）：e12403.
[9] 刘伟江，魏海，运天鹤.基于卷积神经网络的客户信用评估模型研究[J].数据分析与知识发现，2020，4（6）：80-90.
LIU W J，WEI H，YUN T H.Evaluation model for customer credits based on convolutional neural network[J].Data Analysis and Knowledge Discovery，2020，4（6）：80-90.
[10] 牛晓健，凌飞.基于组合学习的个人信用风险评估模型研究[J].复旦学报（自然科学版），2021，60（6）：703-719.
NIU X J，LING F.Study on personal credit risk assessment model based on hybird learing[J].Journal of Fudan University（Natural Science），2021，60（6）：703-719.
[11] WANG D，LIN J，CUI P，et al.A semi-supervised graph attentive network for financial fraud detection[C]//2019 IEEE International Conference on Data Mining（ICDM），2019：598-607.
[12] LEE J W，LEE W K，SOHN S Y.Graph convolutional network-based credit default prediction utilizing three types of virtual distances among borrowers[J].Expert Systems with Applications，2021，168：114411.
[13] 李思迪，郭炳晖，杨小博.基于图神经网络的金融征信研究[J].计算机科学，2021，48：85-90.
LI S D，GUO B H，YANG X B.Study on financial credit information based on graph neural network[J].Computer Science，2021，48（4）：85-90.
[14] WANG D，ZHANG Z，ZHOU J，et al.Temporal-aware graph neural network for credit risk prediction[C]//Proceedings of the 2021 SIAM International Conference on Data Mining（SDM），2021：702-710.
[15] 马帅，刘建伟，左信.图神经网络综述[J].计算机研究与发展，2022，59（1）：47-80.
MA S，LIU J W，ZUO X.Survey on graph neural network[J].Journal of Computer Research and Development，2022，59（1）：47-80.
[16] BRUNA J，ZAREMBA W，SZLAM A，et al.Spectral networks and deep locally connected networks on graphs[C]//2nd International Conference on Learning Representations，2014.
[17] HAMILTON W，YING Z，LESKOVEC J.Inductive representation learning on large graphs[C]//Advances in Neural Information Processing Systems，2017.
[18] VELI?KOVI? P，CUCURULL G，CASANOVA A，et al.Graph attention networks[C]//Proceedings of the International Conference on Learning Representations，2018.
[19] VASWANI A，SHAZEER N，PARMAR N，et al.Attention is all you need[C]//Advances in Neural Information Processing Systems，2017：5998-6008.
[20] WIGINTON J C.A note on the comparison of logit and discriminant models of consumer credit behavior[J].Journal of Financial Quantitative Analysis，1980，15（3）：757-770.
[21] CORTES C，VAPNIK V.Support-vector networks[J].Machine Learning，1995，20（3）：273-297.
[22] CHEN T，GUESTRIN C.XGBoost：a scalable tree boosting system[C]//Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining，2016：785-794.
[23] KE G，MENG Q，FINLEY T，et al.LightGBM：a highly efficient gradient boosting decision tree[C]//Advances in Neural Information Processing Systems，2017.
[24] BREIMAN L.Random forests[J].Machine Learning，2001，45（1）：5-32.
[25] WOLPERT D H.Stacked generalization[J].Neural Networks，1992，5（2）：241-259.
[26] KRIZHEVSKY A，SUTSKEVER I，HINTON G E.Image-Net classification with deep convolutional neural networks[C]//Advances in Neural Information Processing Systems，2012.
[27] YING Z，BOURGEOIS D，YOU J，et al.GNNExplainer：generating explanations for graph neural networks[C]//Advances in Neural Information Processing Systems，2019.