基于CatBoost算法在P2P借贷信用风险的研究
Research on Credit Risk of P2P Lending Based on CatBoost Algorithm

DOI: 10.12677/FIN.2019.93015, PP. 137-141

Keywords: CatBoost，信用评分，机器学习
CatBoost, Credit Risk, Machine Learning

Full-Text   Cite this paper   Add to My Lib

Abstract:

在互联网金融和数据挖掘技术的发展下，运用机器学习算法在传统金融领域和P2P平台领域，对降低借款人的违约风险，保证金融行业与P2P平台良好运营具有重要意义。本文利用澳大利亚P2P平台Ratesetter官网上的贷款数据，通过CatBoost算法与传统机器学习算法作比较分析，以AUC值和准确率作为评价标准，实证研究显示CatBoost算法在对信用评分中优于传统机器学习算法，能够达到更好的准确性。
Under the development of Internet finance and data mining technology, the use of machine learning algorithms in the traditional financial field and P2P platform field is of great significance to reduce the default risk of borrowers and ensure the good operation of financial industry and P2P platform. This paper uses the loan data of Australia P2P platform, compares with CatBoost algorithm and traditional machine learning algorithm, and uses AUC value and accuracy as evaluation standard. Empirical research shows that CatBoost algorithm is superior to traditional machine learning algorithm in credit scoring and can achieve better accuracy.

References

[1]	秦宛顺, 石庆焱. 一个基于Logistic回归的个人信用评分模型[J]. 21世纪数量经济学(第4卷), 2003.
[2]	刘云焘, 吴冲, 王敏, 等. 基于支持向量机的商业银行信用风险评估模型研究[J]. 预测, 2005, 24(1): 52-55.
[3]	张道宏, 张璇, 尹成果. 基于BP神经网络的个人信用评估模型[J]. 情报杂志, 2006, 25(3): 68-70.
[4]	杨海江, 魏秋萍, 张景肖. 基于改进的AdaBoost算法的信用评分模型[J]. 统计与信息论坛, 2011, 26(2): 27-31.
[5]	蒋翠清, 王睿雅, 丁勇. 融入软信息的P2P网络借贷违约预测方法[J]. 中国管理科学, 2017, 25(11): 12-21.
[6]	谭中明, 谢坤, 彭耀鹏. 基于梯度提升决策树模型的P2P网贷借款人信用风险评测研究[J]. 软科学, 2018(12): 136-140.
[7]	Dorogush, A.V., Ershov, V. and Gulin, A. (2018) CatBoost: Gradient Boosting with Categorical Features Support. arXiv:1810.11363
[8]	Nguyen, V.K., Zhang, W.E. and Sheng, Q.Z. (2018) Identifying Price Index Classes for Electricity Consumers via Dynamic Gradient Boosting. In: Hacid, H., Cellary, W., Wang, H., Paik, H.Y. and Zhou, R., Eds., Web Information Systems Engineering WISE 2018, WISE 2018. Lecture Notes in Computer Science, Vol. 11234. Springer, Cham, 472-486. https://doi.org/10.1007/978-3-030-02925-8_33

Full-Text