Research On Identifying Urban Regions Of Different Functions From WeChat Data And POIs

There are numerous data with different kinds and structures constantly being produced from every corner of urban.These data could be perception and storage very well with the rapid development of science and technology.How to effectively use the data produced,manage and arrange the heterogeneous data,and analyze the data by using technical means to provide services for the future city,have been important research topics for the scholars.Urban planning is one class of the hot topic-urban computing.And urban functional regions identification is the basis of urban planning.Understanding urban regions of different functions can let people understand the complex urban spatial structure rapidly,grasp the current urban space solve the problems and make plans.It is also the basis of exploring the composition of different functional urban regions,and understanding the function and distribution of urban regions.In the past,the main data sources used to identify urban regions of different functions are the data of floating car,cellphone details and POIs.However,the floating car data can not accurately locate human activities to regions,and the size of cellphone detail data are not big enough.Accordingly,we propose a method to identify urban regions of different functions by using WeChat data and POIs.The data used in this paper include static data and dynamic data to identify urban regions of different functions which is closer to reality.POI data are seen as static data,and stations' flows,outflows and inflows of people hourly extracted from WeChat data are seen as dynamic data.The main steps of this method are as follows:Firstly,we divide urban space according to the location of cellphone base station.Secondly,we extract the base stations' flow matrix,outflow matrix and inflow matrix of people hourly from WeChat data,and extract the class distribution matrix of POIs from POI data.Then we normalize the four matrices with the factor of base stations' area.Thirdly,we combine the four matrices together as one matrix Y,and delete the outliers of Y to extract the base stations with no outliers.Then we divide the urban space,extract and normalize the matrices again according the base stations with no outliers to generate a new Y for clustering.Lastly,we calculate the overlap ratio of the clustering results and the distribution of POIs with obvious characteristics(such as view spot and residence,etc.)in the city.According to the overlap ratio,we can identify the functions of each base station,in other words,we can identify urban regions of different functions from this.In order to verify the effectiveness of the proposed method,we choose a certain range of Hangzhou as the research object for testing.In this paper,we contrast the result of the test with the real distribution of urban regions and the clustering results of other data sources to evaluate the validity and accuracy of the proposed method.The result shows this proposed method works which can verify that using WeChat data and POI data to identify urban regions of different functions can realize the function division and feature analysis of urban regions accurately,and provide the technical support for urban structure layout,land development and utilization.It is also the realistic basis for reasonable utilization of urban spaces.