| Sun Lingli,Yang Guijun,Wang Yutong.Random Forest Model Based on Benford‘s Law and Its Application in Financial Early Warning[J].The Journal of quantitative and technical economics,2021,(9):159-177 |
| 基于Benford律的随机森林模型及其在财务风险预警的应用 |
| Random Forest Model Based on Benford‘s Law and Its Application in Financial Early Warning |
| |
| DOI： |
| 中文关键词: 财务风险 Benford律 随机森林 |
| 英文关键词: Financial Early Benford‘s Law Random Forest |
| 基金项目:本文获得国家社科基金青年项目“轮换样本校准估计方法在中国住户调查中的应用研究”（20CTJ009）、国家社科基金重点项目“基于大数据的人口统计调查方法与应用研究”（20ATJ008）、天津市统计局统计科研项目“行业间工资差距的演进趋势分析”（TJ2020-2021KY010）、天津市2019年度哲学社会科学规划重点课题“大数据背景下多目标抽样设计的理论和应用”（TJTJ19-001）的资助。 |
|
| Hits: |
| Download times: |
| 中文摘要: |
| 研究目标：财务数据质量对公司财务风险预警至关重要。数据质量差的财务数据往往降低财务风险预警模型的有效性。利用Benford律能有效评价财务数据质量的特点，构建带有Benford因子的随机森林模型，用于处理财务数据质量对财务风险预警模型带来的影响，能有效提高财务预警模型的预测精度。研究方法：通过Benford律检验财务数据质量，构造Benford因子添加到财务指标变量中，建立基于Benford律的随机森林模型。选择中国A股和美国股市上市公司的财务指标数据进行对比实证分析，采用学习曲线对模型进行参数调优确定最终模型，对比基于Benford律随机森林模型和随机森林模型的预测效果。研究发现：Benford因子能够识别存在财务舞弊的具体样本点并提供数据质量有关信息。相比随机森林模型，基于Benford律的随机森林模型可以提高财务风险预警的准确率。研究创新：将Benford律引入随机森林模型，构造Benford因子，提出基于Benford的随机森林模型。研究价值：基于Benford律的随机森林模型具有更高的预测准确率，扩展了随机森林模型的实用性，为上市公司财务风险预警研究提供了新视角。 |
| 英文摘要: |
| Research Objectives：The quality of financial data is of vital importance to the financial risks early warning of the companies.Financial data with poor data quality often reduces the effectiveness of financial risk early warning models.Construct a random Forest model with Benford factor by combining Benford‘s law to effectively evaluate the characteristics of financial data quality.The model can be used to deal with the influence of financial data quality on financial risks early warning model and improve the prediction accuracy of finance early warning model effectively.Research Methods：Testing the quality of financial data by Benford‘s law，constructing Benford factors and adding them to financial indicator variables，and establishing a random forest model based on Benford‘s law.Selecting the financial index data of Chinese A-share and US stock market listed companies to conduct comparative empirical analyses.Using the learning curve to optimize the parameters for the final model，and compare the prediction effects of the random forest model and the other one based on Benford‘s law.Research Findings：The Benford factors can identify specific sample points with financial fraud and provide information about data quality.Compared with the random forest model，the one based on Benford‘s law can improve the accuracy of financial risks early warning.Research Innovations：The Benford‘s law is introduced into the random forest model that the Benford factors are constructed and the random forest model based on Benford is proposed.Research Value：The random forest model based on Benford‘s law has higher prediction accuracy，expands the practicability of the random forest model，and provides a new perspective for the researches of financial risks early warning of listed companies. |
|
View Full Text
View/Add Comment Download reader |
| Close |
