基于Benford-LightGBM上市公司财务风险预警研究

lthough the Chinese government continues to increase the supervision of listed companies, listed companies tend to cover up potential problems by whitewashing financial data for the purpose of beautifying their own business conditions, resulting in frequent financial fraud incidents, and it is necessary to build an accurate and reliable early warning model. Based on Benford\'s law, this paper constructs the Benford quality factor and integrates it into the LightGBM model, constructs the Benford-LightGBM financial risk early warning model, takes 2925 Chinese A-share listed companies as samples, and uses the wrapping method to select the features, and compares the prediction effect of the LightGBM model and the Benford-LightGBM. The results show that the Benford-LightGBM model fused with Benford factor can be used to predict the financial risk of listed companies, among the 878 test set companies, 865 Benford-LightGBM models predict accurately, and 859 LightGBM models predict accurately, and compared with the LightGBM model, the recall rate is increased by 14%, the accuracy is increased by 6%, and the F1 value is increased by 0.11. AUC increased from 0.9485 to 0.9757. This finding not only provides a novel and effective solution for the field of financial risk early warning, but also emphasizes the core position of data quality in the construction and application of complex models, and provides a powerful theoretical reference for government supervision, investor decision-making and corporate self-examination.