基于机器学习算法的信用风险量化模型研究  

Research on Credit Risk Quantification Model Based on Machine Learning Algorithm

作　　者：李沐勋 Li Muxun(Antan Credit Rating Co.,Ltd)

机构地区：[1]安泰信用评级有限责任公司

出　　处：《金融经济》2023年第4期75-83,共9页Finance Economy

摘　　要：传统的信用风险计量模型难以处理高维数据和非线性问题,多具备较严格的假设条件,计算结果常与实际情形存在较大的误差。本文综合考虑影响信用风险的内生变量和外生变量,使用更优的非线性变换方式拟合数据,并借助机器学习强大的算力和学习迭代优势量化信用风险。实证结果表明,该模型算法可提高预测结果的拟合度和准确性。Traditional credit-risk measurement models have difficulty in dealing with high-dimensional data and nonlinear problems,and often have strict assumptions,leading to large errors between the calculated results and the actual situation.This paper considers both endogenous and exogenous variables that affect credit risk,uses a more optimal nonlinear transformation method to fit the data,and quantifies credit risk with the powerful computational and iterative learning advantages of machine learning.Empirical results show that the algorithm of this model can improve the fitting and accuracy of predictive results.

关 键 词：债券市场 机器学习 信用风险计量 

分 类 号：F832.5[经济管理—金融学]