基于随机森林的债券违约预测  

Prediction of Bond Default Based on Stochastic Forest

作　　者：尹涛 李秋敏 YIN Tao;LI Qiu-min(Chengdu University of Information Engineering,Chengdu 610100,China)

机构地区：[1]成都信息工程大学,成都610100

出　　处：《价值工程》2023年第1期120-122,共3页Value Engineering

摘　　要：本文利用方差选择法和互信息法抽取了财务特征中与债券违约最相关的特征,然后使用SMOTE和Tomek Links结合的方法做样本平衡处理,最后构建基于随机森林算法的债券违约预测模型,并与逻辑回归、决策树构建的模型进行了预测性能上的对比。结果显示,基于随机森林构建的模型相比于逻辑回归、决策树构建的模型,AUC、准确率这两个指标的值都更高,表明该研究在债券违约预测方面具有一定的参考价值。In this paper,the variance selection method and mutual information method are used to extract the financial features that are most relevant to bond default,and then the combination of SMOTE and Tomek Links is used to balance the samples.The models constructed by logistic regression and decision tree were compared in terms of predictive performance.The results show that the model based on random forest has higher values of AUC and accuracy than the model built by logistic regression and decision tree,indicating that the research has certain reference value in bond default prediction.

关 键 词：方差选择法 互信息法 违约预测 随机森林 

分 类 号：F832.5[经济管理—金融学]