关键词:
债券违约
财务指标
线性判别分析
支持向量机
摘要:
近年来,债券市场已不再是稳固收益。随着第一支债券违约发生,债券市场打破刚性兑付情况。而债券违约预测成为众多学者关注重点。本文以发行债券的150家企业为研究对象,通过分析各企业一年期的财务指标,反应企业偿债能力、盈利能力等多个方面。以fisher线性判别分析(linear discriminant analysis, LDA)进行降维,构建了LDA-SVM模型对债券违约与否进行预测。并对比了逻辑回归(logit)、支持向量机(Support Vector Machine, SVM)、XGboost等二分类预测模型,其结果表明本文模型效果显著,有90%的准确率,对债券违约预测提供了有效思路,为投资人对于债券违约风险提供了参考。In recent years, the bond market has ceased to be a solid yield. With the default of the first bond, the bond market broke the rigid payment situation. The prediction of bond default has become the focus of many scholars. This paper takes 150 companies that issued bonds as the research object, and analyzes the one-year financial indicators of each enterprise to reflect the solvency and profitability of enterprises. Fisher linear discriminant analysis (LDA) was used to reduce dimensionality, and the LDA-SVM model was constructed to predict whether the bond would default or not. The study compared several binary classification prediction models, including Logistic Regression (logit), Support Vector Machine (SVM), and XGBoost. The results showed that the model proposed in this paper performed significantly well, achieving an accuracy of 90%. This provides an effective approach for predicting bond defaults and offers valuable insights for investors in assessing bond default risks.