Points Of Interest Detection And Spatial Interaction Analysis Based On The Trajectory Of Taxi And Ride-hailing

The point of interest(POI)is a geographic entity with specific functions that can be abstracted into a point element.The POI is strongly related to origin and destination especially in the field of public transportation,which can be applied to the analysis of urban travel preferences.Traditionally,floating cars are important to the transportation in cities,mainly including taxis and buses.The trajectory of the floating car reflects the spatial interaction process between geographical entities and can be used as the basis for the study of the spatial-temporal laws.Since taxis have a "one-stop" operation model,the study of the origin and destination can be transformed into a study on a pair of related POIs.In recent years,online ride-hailing service has become more popular,providing residents with more choices for travel.Due to the obvious similarities between taxis and ride-hailing service,in order to avoid excessive and improper competition,it is necessary to analyze the evolutionary law and relationship between taxis and ride-hailing service from the perspective of spatial interaction.In this paper,we used the trajectories of taxis and ride-hailing service in Wuhan in 2018 as the main data source,and proposed a travel POI detection method based on models such as address trees and support vector machine(SVM).In addition,we introduced complex network theory and built the multi-layer spatial interaction network based on the travel POI of taxi and ride-hailing.Subsequently,based on methods such as time series analysis and correlation analysis,we studied the structural features and correlations of the spatial interaction network to obtain the spatial interaction law between taxis and ride-hailing.The main work of this paper can be summarized as follows:(1)Study on travel POI detection method based on the trajectory of taxi and ride-hailingThe origin and destination(OD)points of the trajectory contain important temporal and spatial information and attribute information,which can be transformed into POIs that are attractive to passengers.Based on the address tree model,we first processed the pick-up and drop-off addresses in the trajectory according to a unified standard to extract the address elements and the affiliation between the elements.Then,based on the SVM model,we matched the address elements to obtain the travel POI.For the taxi trajectory,we extracted the affiliation between the OD points and the road based on the map matching method,and used the Jenks method to determine the appropriate thresholds to establish the association between the taxi trajectory and the travel POI.(2)Analysis of Evolution Law and Correlation Relationship of Spatial Interaction Network Based on Travel POIs of Taxi and Ride-hailingComplex network is an effective way to express the relationship and interaction mode between individuals in the complex systems.We used the travel POIs as nodes and convert trajectories to edges to form the spatial interaction networks of taxi and ride-hailing.Then we comprehensively used the spatial-temporal analysis methods such as structural feature measurement methods,time series analysis methods,and spatial overlay analysis related to complex networks to explore the evolution laws of the internal structure of the spatial interaction networks of taxis and ride-hailing.In addition,we conducted a correlation analysis of the inter-layer structure and attribute features of the multiple spatial interaction networks formed by taxis and ride-hailing to supplement the spatial interaction laws of taxi and ride-hailing travel from the perspective of the interaction between complex systems.Therefore,the analysis of spatial interaction laws,as a continuation of the study on travel POI detection,reflects the social spatial characteristics of the city,which can support the rational dispatch of urban taxis and ride-hailing service and the optimal allocation of public resources.