基于决策树算法的上市公司股东行为研究

发布时间：2018-03-13 18:28

本文选题：统计分类　切入点：股东行为　出处：《数理统计与管理》2017年01期 　论文类型：期刊论文

【摘要】：本文从生态学视角出发,将研究焦点深入到上市公司内部的股东层面。通过将上市公司股东进行生态学的种群划分,刻画不同类型股东所处的生态位,研究分析不同生态位的股东群体之间的关系和行为表现。基于国泰安数据库和CCER数据库2643家上市公司股东2008-2014年四个季度约60万条记录数据,将决策树算法用于股东统计分类和行为分析上,尝试了一种具有速度快、精确度高的特点的统计分析方法;通过分析发现股东持股排名的生态位维度对股东增减持行为和程度影响最大,为进一步的治理决策提供了依据。
[Abstract]:From the perspective of ecology, this paper focuses on the shareholders of listed companies. By dividing the shareholders of listed companies into ecological groups, this paper depicts the niche of different types of shareholders. Based on Cathay Pacific database and CCER database, there are about 600,000 records of shareholders in four quarters from 2008 to 2014. The decision tree algorithm is applied to the statistical classification and behavior analysis of shareholders, and a fast and accurate statistical analysis method is tried. Through the analysis, it is found that the niche dimension of shareholder ownership ranking has the greatest influence on the behavior and degree of shareholders' increase and decrease, which provides the basis for further governance decisions.
【作者单位】：北京工业大学经管学院;北京工业大学数理学院;
【基金】：国家自然科学基金项目(11372014,11072007) 国家社会科学基金重点项目(13AGL003) 教育部人文社科基金项目(10YJA630185) 北京市自然科学基金项目(1122001)
【分类号】：F832.51

【参考文献】