Spatial Correlation Analysis Of Node Centrality Of Directed Weighted Geographic Networks Constructed From Telecommunication Data Of Milan,Italy

The rapid development of communication technology has spawned a huge amount of telecommunication data with location information.Using big data technology to analyze location-based telecommunication data,we can understand the dynamic changes of communication interaction modes between different regions.By correlating the communication interaction mode with the features in the relevant geospatial,we can further understand the adaptability between the crowd activity and the urban spatial structure,thus providing auxiliary decision-making for urban sustainable development and optimization of urban management.Understanding the communication interaction mode between different regions from the perspective of the network can grasp the law of the interaction mode from a more macroscopic and holistic perspective.The analysis of the autocorrelation of the centrality of the nodes in the network and the correlation between the centrality of the nodes and the features of the corresponding spatial regions of the nodes are the main methods for analysis from the network perspective.This paper proposes a spatial correlation analysis of the centrality of directed weighted geographic network nodes in telecommunication data.Specifically,it includes:(1)Using the Spark big data platform to construct a directed weighted geographic network from the telecommunication data,and calculate the node centrality value.(2)By constructing a thematic map of multi-class nodes,calculating the Moran index value of the centrality of node,the global spatial correlation of node centrality in directed weighted geographic networks is analyzed.(3)Using the multi-period node-centric thematic map and the Moran index value,analyze the law of the overall autocorrelation of the node's centrality with time.(4)Using a variety of clustering analysis methods to analyze the local spatial correlation of node centrality in directed weighted geographic networks.According to the results of cluster analysis,multiple POIs in different regions are further classified and analyzed,and the correlation between node centrality and feature attributes is analyzed.This paper selects the city of Milan,Italy as the research area,and analyzes the telecommunication data and 9 types of POI data between the regions in the region from November 1 to December 15,2013.The experimental results show that:(1)The centrality of the nodes of the directed weighted geographic network in telecommunication data is positively correlated as a whole,and the nodes in the southwestern region have the highest centrality and high-value aggregation distribution.The northwest and southeast are second to the southwestern,and the northeast region is the lowest.(2)On working days,people tend to be relatively dispersed,and the spatial correlation of node centrality is relatively low;on weekends,people will concentrate from various areas to key areas,and the spatial correlation of node centrality is relatively strong.(3)The hot spots are mainly concentrated in the southwestern part of Milan,and the cold spots are mainly concentrated in the northeast region.For the nine POI data,the hotspot area is on average higher than the cold spot area.