主管:中国科学院
主办:中国优选法统筹法与经济数学研究会
   中国科学院科技战略咨询研究院

中国管理科学 ›› 2026, Vol. 34 ›› Issue (10): 107-117.doi: 10.16381/j.cnki.issn1003-207x.2023.1157

• • 上一篇    

大数据背景下DEA方法的加速算法研究

朱庆缘1,2, 徐树奇3(), 潘应浩4, 吴杰4   

  1. 1.南京航空航天大学经济与管理学院,江苏 南京 211106
    2.南京航空航天大学能源软科学研究中心,江苏 南京 211106
    3.合肥大学管理学院,安徽 合肥 230601
    4.中国科学技术大学管理学院,安徽 合肥 230026
  • 收稿日期:2023-07-11 修回日期:2024-03-26 出版日期:2026-10-25 发布日期:2026-10-09
  • 通讯作者: 徐树奇 E-mail:xusq@hfuu.edu.cn
  • 基金资助:
    国家自然科学基金项目(71904084);国家自然科学基金项目(72271121);国家自然科学基金项目(72371232);国家自然科学基金项目(71921001);国家自然科学基金项目(71971203);国家自然科学基金青年项目(724B2031);中国博士后科学基金特别资助项目(2020TQ0145);中央高校基本科研业务费专项资金项目(WK2040000027);合肥大学人才科研基金项目

The Computation Acceleration of DEA in the Big Data Context

Qingyuan Zhu1,2, Shuqi Xu3(), Yinghao Pan4, Jie Wu4   

  1. 1.College of Economics and Management,Nanjing University of Aeronautics and Astronautics,Nanjing 211106,China
    2.Research Center for Energy Soft Science,Nanjing University of Aeronautics and Astronautics,Nanjing 211106,China
    3.School of Management,Hefei University,Hefei 230601,China
    4.School of Management,University of Science and Technology of China,Hefei 230026,China
  • Received:2023-07-11 Revised:2024-03-26 Online:2026-10-25 Published:2026-10-09
  • Contact: Shuqi Xu E-mail:xusq@hfuu.edu.cn

摘要:

数据驱动的多属性非参数理论——数据包络分析(DEA)方法,已在绩效评价、资源分配、成本分析等领域得到广泛应用,并成为运营管理领域的重要方法之一。近年来,随着现代社会飞速运转所产生的数据种类及规模的爆发式增长,原本初显成熟的DEA方法遭遇了全新挑战,如在大数据背景下,随着指标、决策单元的数量增长,以及数据复杂度的提高,DEA方法所需的计算时间激增。在系统综述现有DEA计算加速理论方法及其优劣势的基础上,本文提出了一种具有优越计算性能的单机单线程DEA方法的加速算法,该算法能够大幅节省DEA方法的计算时间,并被证明具有优越的数据清洗能力。应用于我国A股上市公司2016-2021年的财报数据组以及两个仿真数据组,以验证其数据清洗能力及其在实践中的更多价值。

关键词: 数据包络分析, 大数据, 计算加速

Abstract:

Data envelopment analysis (DEA), a data-driven, multi-attribute, nonparametric approach, has been widely applied in performance evaluation, resource allocation, cost analysis, and related fields, and has become an important methodology in operations management. In recent years, the explosive growth in the variety and scale of data generated by the rapid development of modern society has posed new challenges to the DEA methodology, which had previously reached a relatively mature stage. In the context of big data, the increasing number of indicators and decision-making units, together with rising data complexity, has led to a sharp increase in the computational time required by DEA models. Based on a systematic review of existing theoretical and methodological approaches for accelerating DEA computation, as well as their respective advantages and limitations, this study proposes a computationally efficient acceleration algorithm for single-machine, single-thread DEA implementation. The proposed algorithm can substantially reduce the computational time of DEA models and is shown to possess strong data-cleaning capability. The algorithm is applied to real financial statement data from Chinese A-share listed companies from 2016 to 2021, as well as two simulated datasets, to validate its data-cleaning capability and further demonstrate its practical value.

Key words: data envelopment analysis, big data, computation acceleration

中图分类号: