Taxi GPS Data Analysis Model Based On Complex Network And Dynamic Network Anomaly Detection

In this study,an analytical framework of taxi GPS data set was proposed,which was stud ied from three aspects:spatial and temporal characteristics of urban traffic travel and abnorma 1 detection of urban traffic dynamic network,and an empirical study was conducted using GP S data of New York City taxis.The research of this paper provides a feasible research framew ork for urban traffic researchers.It provides data support and theoretical support for New Yor k City’s urban planning and traffic decision-making.Provide guidance for taxi drivers in plann ing routes and areas;Planning travel plans for citizens to avoid traffic hotspots and hot spots t o provide guidance.The research of this paper includes the following aspects:Firstly,this paper introduces the research status of complex network,urban road traffic a nomaly detection and taxi GPS data.Based on this,a taxi GPS data analysis model based on c omplex network and dynamic network anomaly detection is proposed.The model includes tw o parts:the establishment and topological property analysis of urban travel network based on t axi GPS data and traffic anomaly detection based on SpotLight and Isolation Forest.The first part uses the spatial data of taxi GPS data set to cluster,so as to divide the traffic area.Traffic areas are abstracted as nodes,trips between traffic areas are abstracted as edges,and the numb er of trips is abstracted as weights to establish an urban travel network.In the second part,the urban traffic travel network is established respectively per hour to form a dynamic network of urban traffic travel.The dynamic network is mapped to the space by SpotLight algorithm of d ynamic network anomalies,and the anomaly detection is carried out by Isolaiton Forest algori thm to discover traffic anomalies.Secondly,the GPS data set of New York City taxis is introduced and analyzed.In order t o better understand the results of the follow-up experiment,the geography and traffic status of New York City and the New York City geographic information database,which provides the urban geographic information for this experiment,are first introduced.The New York City ta xi data set includes 92 days of data from November 1,2015 to January 31,2016.A total of 32,675,244 trip samples were obtained after data cleaning and data screening,such as longitude a nd latitude screening,eliminating the trip records where null values and outliers were located.The distribution map of taxi daily trips in New York City was drawn,and the hourly distributi on map of traffic trips in working days,weekends and holidays was analyzed.The taxi trips in New York City were explored from the time aspect.Draw the spatial distribution map of start ing point finishing point,starting point and finishing point of taxi travel in New York City.Finally,this paper empirically analyzes the taxi GPS data analysis model based on compl ex network and dynamic network anomaly detection in New York City taxi trip data set.First,combined with the geographical characteristics of New York City,this paper divides the traff ic area of New York City based on the K-means clustering algorithm,establishes the urban tra ffic network of New York City,and calculates the topological properties such as degree distri bution and average network diameter.Then,the dynamic network anomaly detection method based on SpotLight and Isoaltion Forest is used to detect traffic anomalies,and the anomaly d etection results are analyzed in combination with real world information.