摘要
在国际评级机构经济背景与我国市场经济不相契合的前提下,为完善我国上市公司信用评级方法体系,本文通过完善指标选取、数据处理与算法优化三个方面,选取锐思数据库我国上市公司的评级数据、财务数据与相关非财务数据,采用"数据分割"方式,创新性地将人工蜂群算法优化支持向量机(ABC-SVM)应用于我国上市公司信用评级中。结果表明:通过将整体数据集分割为升级数据集和降级数据集,评级准确率分别提高了3.78%和3.37%;同时,较传统支持向量机与其他两种生物启发算法(粒子群算法、遗传算法)优化下的支持向量机,ABC-SVM算法的评级效果最好,评级准确率显著提高了5%—7%。本文为我国上市公司信用评级提供了方法思路,丰富了企业信用评级指标体系,为建立层次更全面的企业信用数据库、提高我国上市公司信用评级准确率提供技术支持,并为我国争取国际评级话语权提供理论依据。
In order to improve the credit rating system of listed companies in China,this paper selects the rating data,financial data and relevant non-financial data of listed companies from RESETIS database and innovatively applies the artificial swarm algorithm optimization support vector machine(ABC-SVM)to the credit rating of listed companies in China by improving the three aspects of indicator selection,data processing and algorithm optimization.The results show that after segmenting the overall data set into an upgraded data set and a downgraded data set,the rating accuracy rate increased by 3.78%and 3.37%,respectively.At the same time,compared with the traditional support vector machine and other two kinds of biological heuristic algorithm(particle swarm algorithm,genetic algorithm),the ABC-SVM algorithm has the best rating effect,and the rating accuracy rate is significantly improved by 5%-7%.This paper provides a new method for the credit rating of listed companies in China,enriches the enterprise credit rating index system,and provides a theoretical basis for the establishment of a more comprehensive enterprise credit database and provides a theoretical basis for Chinese enterprises to strive for the right to international rating discourse.
作者
马晓君
宋嫣琦
张萌
MA Xiao-jun;SONG Yan-qi;ZHANG Meng(School of Statistics,Dongbei University of Finance and Economics,Dalian 116025,China)
出处
《东北财经大学学报》
2020年第4期57-65,共9页
Journal of Dongbei University of Finance and Economics
关键词
人工蜂群优化算法
支持向量机
数据分割
信用评级
上市公司
artificial bee colony optimization algorithm
support vector machine
data segmentation
credit rating
listed company
