GDP Spatial Differentiation in the Perspective of Urban Functional Zones

Abstract

The current spatial analysis of urban GDP mainly focuses on macroscopic scales such as city and county level. However, inadequate attention has been paid to the fine-grained spatial units within cities. This study aims to fill this gap by exploring the issue of GDP spatial differentiation from the Urban Functional Zone (UFZ) perspective. First, the road network was used to delineate urban units, and UFZs were identified using XGBoost classifier with multi-source data. Second, the distribution of GDP was analyzed using the spatial autocorrelation model. Third, the correlation between types of UFZs and GDP spatial cluster was explored using random forest algorithm. Results indicate that: (1) the UFZs can be identified with an overall precision of 0.931, and the industrial and residential are the dominated functions in the study area; (2) The GDP are significantly different from the UFZs in the urban center to the zones in periphery of the study area and have a strong positive spatial autocorrelation; (3) Among all the UFZs, residential have the strongest correlation with the GDP spatial clusters.  This study contributes to urban planning, construction, and coordinated regional development by providing replicable methods and ideas for adjusting and optimizing urban industrial structure.

Keywords: Urban functional zone, GDP, Multi-source data, XGBoost classifier, Random Forest