基于POI数据的特大城市商业业态空间分布及影响因素差异研究——以广州为例

Study on the Spatial Distribution of Mega-City Commercial Formats and the Differences of Influencing Factors Based on POI Data—A Case Study of Guangzhou

摘要: 选取广州为研究案例地，基于多类商业网点POI数据，利用空间核密度估计、DBSCAN聚类、负二项回归等方法探讨特大城市商业空间总体布局及细分业态的分布特征和影响因素差异。结果表明：1) 广州城市商业空间布局呈现中心城区高度集聚的多中心特点。2) 不同商业业态的空间集聚特征及核密度值均存在明显差异，其中专业店专卖店核密度值最大且呈现“多点集聚，分散分布”特征，购物中心核密度值较小且呈现“双核集聚，多点扩散”特征，商厦呈现“单核集聚，圈层扩散”特征，超市呈现“中心城区连片分布，外围区域分散分布”特征，便利店呈现“多点集聚，网络分布”特征，连锁便利店在主城区集聚分布，家居建材市场呈现近郊区集聚分布特征。3) 负二项回归模型估计结果表明人口规模对各商业业态的影响均显著为正；交通可达性对商业业态网点分布总体呈现正向影响，道路密度对大型超市、超市、购物中心、商厦、家居建材市场等商业业态分布表现为正向影响，地铁站点对购物中心、商厦等商业业态分布呈现正向影响；商务办公区与商业业态布局呈现显著正相关，住宅功能区与商业业态布局呈现负相关，住宅功能区与商业设施存在空间分离的趋势；区位因素仅与购物中心、商厦等商业业态存在显著负相关，表明广州购物中心、商厦分布存在朝外围区域集聚的趋势。

Abstract: In this paper, we explore the spatial distribution characteristics of mega-city commercial formats and the differences of influencing factors. Guangzhou is selected as the location of case study and Point of Interest (POI) data of various commercial facilities is used in combination with spatial kernel density analysis, DBSCAN spatial clustering and negative binomial regression method. Our results indicate that: 1) The layout of urban commercial space in Guangzhou forms a multi-central agglomeration, and the degree of agglomeration in downtown is higher. 2) There are differences in the spatial agglomeration characteristics and core density of different commercial formats. The core density of specialty stores is the largest and presents the characteristics of “multi-point agglomeration and decentralized distribution”. The core density of the shopping center is small and presents the characteristics of “dual-core agglomeration and multi-point diffusion”. Commercial buildings are characterized by “single-core agglomeration and circle diffusion”. Supermarkets are characterized by “continuous distribution in the central urban area and scattered distribution in the peripheral areas”. Convenience stores are characterized by “multi-point agglomeration and network distribution” and the convenience store chain is distributed mainly in the city center. Homebuilding materials markets are distributed in the suburbs. 3) Regarding the estimation results of negative binomial regression model, population has a significant positive impact on the distribution of all commercial formats. Traffic accessibility has a positive impact on the distribution of all commercial formats. Road density has a positive impact on the distribution of commercial formats such as large supermarkets, supermarkets, shopping centers, commercial buildings, home building materials markets and other commercial formats. Subway stations have a significant impact on the distribution of shopping malls, commercial buildings and other commercial formats. There is a significant positive correlation between business office area and commercial formats, a negative correlation between residential functional area and commercial formats, and there is a trend of spatial separation between residential functional areas and commercial facilities. Location factors are a significant negative effect on the distribution of shopping malls, commercial buildings and other commercial formats. This shows that shopping malls and commercial buildings in Guangzhou tend to cluster toward peripheral areas.