基于文本挖掘和二元Logistics回归的生物医药行业上市公司财务预警研究
Research on Financial Early Warning of Listed Companies in Biopharmaceutical Industry Based on Text Mining and Binary Logistic Regression

本研究论文主要探讨了基于文本挖掘和二元logistics回归的生物医药行业上市公司财务预警研究。通过分析生物医药行业上市公司的财务数据和管理层讨论与分析(MD&A)文本，利用二元logistics回归模型和文本挖掘技术，研究构建了一套财务预警模型，以预测企业是否可能发生财务危机。研究选取了现金比率、现金流量负债比率、资产负债率、净资产收益率、营业毛利率、息税前营业利润率、总资产增长率、营业收入增长率、资本保值增率、每股经营活动产生的净流量增长率等财务指标，以及基于MD&A文本信息构建的语调指标作为模型的关键预警指标。这些指标涵盖了企业的多个财务维度，旨在全面评估公司的财务健康状况。模型在非ST公司的判断准确率达到了87.50%，在ST公司的判断准确率为80.00%，整体准确率为85.29%，显示了该模型对于早期识别财务风险具有一定的有效性。
This research paper mainly discusses the research on financial early warning of listed companies in the biomedical industry based on text mining and binary logistic regression. By analyzing the financial data and management discussion and analysis (MD&A) texts of listed companies in the biopharmaceutical industry, and using binary logistic regression models and text mining technology, a set of financial early warning models were constructed to predict whether a company may have a financial crisis. The study selected cash ratio, cash flow-liability ratio, asset-liability ratio, return on net assets, operating gross profit margin, operating profit margin before interest and taxes, total asset growth rate, operating income growth rate, capital preservation growth rate, and operating activities per share. Financial indicators such as the generated net traffic growth rate, as well as tone indicators constructed based on MD&A text information are used as key early warning indicators of the model. These indicators cover multiple financial dimensions of a business and are designed to provide a comprehensive assessment of a company’s financial health. The model’s judgment accuracy in non-ST companies reached 87.50%, in ST companies it was 80.00%, and the overall accuracy was 85.29%, showing that the model has certain effectiveness in early identification of financial risks.