Modeling of regional economic spatial correlation and influence mechanism based on fusion graph neural network

View at publisher Summarise with VIRAT
Authors: Deli Dong, Xuebin Huang, Xiaoshuo Zhao, Fangyuan Chen, Dong Wang
Cited by: 0
References: 15

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Aiming at the shortcomings of traditional regional economic analysis in capturing complex nonlinear spatial correlations, this study aims to construct a model that fuses multi-source big data and graph neural network to quantify global regional economic spatial correlations and analyze its influence mechanism. This study integrates multi-dimensional data such as information flow, capital flow and population flow to construct a global urban economic network, adopting a hybrid model of graph convolutional network and graph attention network, and combining it with a spatial econometric analysis framework. The results show that the determination coefficient (R²) of this model in the economic growth forecast of 127 cities around the world is as high as 0.921, which is significantly better than the traditional spatial econometric model. In the robustness test, the model R2 remains at 0.901 even if 10% of the network connections are randomly deleted. The conclusion proves that the framework can effectively reveal the nonlinear characteristics and multi-scale influence mechanism of regional economic correlations, especially the attention mechanism plays a key role in distinguishing heterogeneous spatial influences, which provides a scientific basis for regional coordinated development policies.

Fields of Study & Subjects

Spatial and Panel Data Analysis Regional Economic Trends and Sectoral Structure Agglomeration and Urban Systems

Publication details

Journal / source
International Conference on Engineering Management, Computer Applications, and Supply Chain (EMCAS)
Pages
25–25
Publication date
1 April 2026
Language
EN