基于新闻文本的上市公司财务困境组合预测模型

已有上市公司财务困境预测模型主要是基于结构化数据进行研究，为进一步提高上市公司财务困境预测模型准确率，本文将非结构化数据引入上市公司财务困境预测问题中，研究了基于新闻文本分类的上市公司财务困境预测模型，结合新闻文本信息和财务信息提出上市公司财务困境组合预测模型。本文首先将新闻数据进行预处理，然后基于新闻文本数据通过支持向量机（SVM）进行财务困境预测，同时基于财务数据通过Logistic模型进行财务困境预测，最后采用阈值表决集成策略整合两种模型的预测结果，实验结果证明了模型的有效性。

Combination Financial Distress Prediction Model based on News Text

Existing corporate financial distress prediction model is based mainly on structured data , this paper introduced unstructured data to corporate financial distress prediction model and studied model based on news text. In order to improve corporate financial distress prediction model accuracy, proposed combination prediction model binding text news and financial data company listed. Firstly, the paper proposed financial distress prediction model by SVM based on the news text, and then proposed financial distress prediction model by Logistic based on financial data, finally, integrated predictions of the two models through threshold vote. Experimental results show that the model is effective.