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 (4): 34-46.doi: 10.16381/j.cnki.issn1003-207x.2024.0854cstr: 32146.14.j.cnki.issn1003-207x.2024.0854

• • 上一篇    下一篇

考虑空间邻近效应的经济时间序列预测建模与应用研究

丁松1,2,3(), 神兴傲4, 党耀国5, 郭旭鹏5   

  1. 1.浙江财经大学浙商资本研究院,浙江 杭州 310018
    2.浙江省“八八战略”研究院,浙江 杭州 310018
    3.浙江财经大学经济学院,浙江 杭州 310018
    4.浙江财经大学-中国社会科学院大学浙江研究院,浙江 杭州 310018
    5.南京航空航天大学经济与管理学院,江苏 南京 211106
  • 收稿日期:2024-05-30 修回日期:2024-10-17 出版日期:2026-04-25 发布日期:2026-03-27
  • 通讯作者: 丁松 E-mail:dingsong1129@163.com
  • 基金资助:
    国家自然科学基金项目(71901191);国家自然科学基金项目(72271120);国家社会科学基金重大项目(23&ZD102);浙江省统计科学研究项目(25TJZZ12);杭州市哲学社会科学重点研究基地ESG与可持续发展研究中心项目(25JD053)

Modelling and Application of Economic Time Series Forecasting Considering Spatial Proximity Effect

Song Ding1,2,3(), Xingao Shen4, Yaoguo Dang5, Xupeng Guo5   

  1. 1.Zheshang Capital Market Research Institute,Zhejiang University of Finance and Economics,Hangzhou 310018,China
    2.Zhejiang Institute of “Eight-Eight” Strategies,Hangzhou 310018,China
    3.School of Economics,Zhejiang University of Finance and Economics,Hangzhou 310018,China
    4.Zhejiang Research Institute of ZUFE-UCASS,Zhejiang University of Finance and Economics,Hangzhou 310018,China
    5.College of Economics and Management,Nanjing University of Aeronautics and Astronautics,Nanjing 211106,China
  • Received:2024-05-30 Revised:2024-10-17 Online:2026-04-25 Published:2026-03-27
  • Contact: Song Ding E-mail:dingsong1129@163.com

摘要:

现实经济系统中,主系统发展趋势易受到邻近区域溢出效应的影响,而当前灰色多变量模型仅考虑关联变量的时间效用,忽视了相邻区域的动态空间邻近效应。因此,本文基于地理距离和经济距离构建出新型综合空间邻近项以衡量空间邻近效应,并据此建立三类考虑空间邻近效应的新型灰色多变量模型。同时,创造性地设计出模型智能算法筛选程序,使用三项算法匹配判断准则,为三类灰色多变量模型选择最佳智能寻优算法,提高模型超参数求解的可靠性、稳定性和可解释性。最后,将三类新型灰色多变量模型应用于上海市经济总量的预测研究,综合利用误差指标、统计检验、蒙特卡洛模拟等方法,验证了所提模型的有效性和稳健性。

关键词: 灰色系统, 灰色多变量模型, 空间邻近效应, 灰色预测

Abstract:

In real-world economic systems, the development trends of major systems are often influenced by spillover effects from neighbouring regions. However, existing grey multivariate models mainly focus on the temporal utility of correlated variables while overlooking the dynamic spatial relationships between neighbouring entities. To address this gap, a novel integrated spatial proximity term is introduced based on geographic and economic distances to measure spatial proximity effects. On this basis, three types of grey multivariate models that consider spatial proximity effects have been developed, namely STGM(1, N, M), STGMC(1, N, M), and STDGM(1, N, M) models. Moreover, the challenge of selecting model hyperparameters is tackled by designing an innovative intelligent algorithm selection framework. This framework introduces three criteria for identifying optimal algorithms, including accuracy verification, statistical testing, and parameter sensitivity analysis, thus enhancing the reliability, stability, and interpretability of the model solutions. The effectiveness and applicability of the spatial proximity term are validated through various methods, such as the HLN test, Monte-Carlo simulations, and probability density analysis, from the perspectives of accuracy and stability. Finally, a case study on the economic prediction of Shanghai demonstrates that the integration of the novel spatial proximity term significantly improves the ability of all three grey multivariate models to capture the spatial proximity effects of neighbouring regions, offering enhanced prediction accuracy, reliability, and stability.

Key words: grey system, grey multivariable models, spatial proximity effects, grey forecasting

中图分类号: