数据与计算发展前沿 ›› 2021, Vol. 3 ›› Issue (6): 151-160.

doi: 10.11871/jfdc.10-1649.2021.06.012

• 技术与应用 • 上一篇    

基于数据挖掘的上市公司违规行为关联分析

徐静()   

  1. 北京联合大学,管理学院,北京 100101
  • 收稿日期:2021-05-15 出版日期:2021-12-20 发布日期:2022-01-26
  • 通讯作者: 徐静
  • 作者简介:徐静,北京联合大学管理学院,副教授,北京大学博士后,美国注册管理会计师,在国内外重要期刊及会议上发表学术论文20余篇,完成专著3部,研究成果曾获丝路书香工程重点翻译项目资助,主要研究方向为监管科技和大数据审计。
    XU Jing is currently a Vice Professor of the School of Manage-ment, Beijing Union University, a postdoctoral fellow at Peking University, a cerified management accountant in the United States. Her current research interests include Regulatory Technology and Big Data Auditing. She has published more than 20 papers in international journals and conferences, completed 3 monographs. Her research output has been funded by the National Key Trans-lation Program, the Silk Road Project. E-mail: gltxj@buu.edu.cn
  • 基金资助:
    北京市社会科学基金规划项目“大数据审计模式下财务报表审计线索发现研究”(21GLB015)

Association Analysis of Violations by Listed Companies Based on Data Mining

XU Jing()   

  1. School of Management, Beijing Union University, Beijing 100101, China
  • Received:2021-05-15 Online:2021-12-20 Published:2022-01-26
  • Contact: XU Jing

RichHTML

21

PDF

73

摘要：

【目的】对上市公司违规行为进行识别并有效防治违规事件的发生,一直是备受关注的议题。【应用背景】基于上市公司大数据及关联规则挖掘算法研究违规行为间的关联关系,是不同于传统研究的新视角,能够为识别和预测上市公司违规行为提供线索,有助于监管部门依据违规行为的关联规则开展案件调查。【方法】选取2000-2020年间因违规受罚的上市公司为样本,运用Apriori和Sequence两种关联规则挖掘算法,分别从静态和动态两个角度,挖掘上市公司违规行为的简单关联规则和序列关联规则。【结果】上市公司违规事件往往不是孤立发生的,一种违规行为可能关联着另一种或多种违规行为;公司运营违法违规作为前项更容易引发信息披露虚假、遗漏或延误等违规,这种前项和后项关系符合违规行为的基本逻辑,具有理论上的合理性。【结论】本研究能够为监管部门提供更加多样化的违规案件调查线索和调查方式,对于推进大数据新型执法具有重要的现实意义。

关键词: 违规行为, 关联规则挖掘, Apriori算法, Sequence算法, 特征识别

Abstract:

[Objective] Identifying violations of listed companies and effectively preventing the occurrence of the violations have always been a topic of concern. [Application background] From new perspectives, study on association rules analysis of violations using the big data of the listed companies and association rule mining algorithms is different from traditional researches. It can provide clues for identifying and predicting violations of listed companies, which helps regulatory authorities to carry out case investigation according to the association rules of violations [Methods] Taking Chinese listed companies punished for violation of rules and regulations from 2000 to 2020 as samples, this paper uses Apriori algorithm and Sequence algorithm to mine the simple association rules and sequential association rules among the violations by Chinese listed companies from static and dynamic perspectives. [Results] The results show that violations by listed companies are not isolated, that is, one kind of violation is often associated with another or more violations. As the former item, operational violations are more likely to lead to wrongful information disclosure, omission or delay, and other violations, while the relationship between the former and the latter items conforms to the basic logic. [Conclusions] More diversified investigation clues and methods for regulatory authorities are provided by the study, which has important practical significance for promoting the new law enforcement in the big data environment.

Key words: violations, association rule mining, apriori algorithm, sequence algorithm, feature recognition