Artificial intelligence industrial agglomerations in Beijing:An spatial pattern study based on improved DBSCAN algorithm
As enterprises are key players in industries,the enterprise development determines the industry develop-ment.The spatial pattern of the enterprises has important guiding significance for the development of the industry and its resource allocation.This paper analyzes the spatial distribution characteristics of fine-grained artificial intelligence(AI)industrial agglomerations in Beijing by using the improved Density-Based Spatial Clustering of Applications with Noise(DBSCAN)algorithm to deal with the geographic location information of AI enterprises registered in the city.In terms of improving DBSCAN algorithm,in the first instance the Minpts parameters are adjusted to two dimensions,the total registered capital of enterprises and the number of enterprises.The number of enterprises is more than 5 and the total registered capital is greater than a certain amount,which are two necessary conditions for the formation of in-dustrial agglomeration area.Then,the enterprises located at the boundary in the cluster are chosen as the geographical boundary points which are connected to form the geographical boundary of the AI industry agglomeration area.This paper focuses on the impact of the total registered capital of enterprises and cluster radius on AI industrial agglomera-tions.Meanwhile,the kernel density estimation method is used to verify that the improved DBSCAN has the advant-age of accurately depicting industrial agglomeration areas and identifying industrial agglomeration areas of different sizes.According to the analysis,AI industrial agglomerations in Beijing are concentrated in six districts of the urban center,showing the distribution of'Driven by the two leading urban districts and spread all over Beijing'.Among the sixteen districts of Beijing,Haidian District and Chaoyang District are at the highest level,where the develop-ment of the AI industry is far ahead,followed by Xicheng District,Dongcheng District,Fengtai District and Changping District.Tongzhou District,Daxing District,Pinggu District,Miyun District,Shijingshan District,Fangshan District,Mentougou District,Huairou District and Shunyi District are at a medium level and Yan-qing District is at the lowest.Besides,the AI industry cluster areas such as Zhongguancun,Shangdi,Xierqi,Wudaokou Wangjing,Guomao and Beijing Economic-Technological Development Zone are accurately loc-ated.It is further concluded that Haidian District's AI research talent and Chaoyang District's original enter-prise base in the information technology field are the main drivers of the outstanding AI development in the two districts.Scarce industrial resources,coupled with limited regional functions and other reasons,have led to low-level AI industrial agglomerations in Yanqing District and other suburbs,indicating that the development of AI industry is relatively slow.The spot research showed that the improved DBSCAN algorithm proposed in this paper is effective and accurate.In order to expand AI-related industries and promote the prosperity of AI economy of Beijing,it is suggested that the government further give play to the value of AI clusters as well as take into account the function of all districts of Beijing when formulating AI industrial policies in the future.
artificial intelligenceindustrial agglomerationimproved DBSCAN algorithmkernel density es-timationGISBeijing
NETL
NSTL
万方数据
