Spatial And Temporal Characteristics Analysis Of Taxi Travel Based On GPS Data

Taxis play an important role in urban traffic.The spatiotemporal characteristics of taxis not only reflect the traffic conditions of a city,but also reflect the travel behaviors of urban residents.The operation law of taxis and the travel behavior of residents are characterized by randomness and uncertainty,and my country has not yet realized the information sharing between taxi drivers and passengers,which often leads to regional supply and demand between taxis and passengers.contradiction.Based on the excavation of the temporal and spatial characteristics of taxi travel,this paper deeply discusses the behavioral laws of taxi operation,so as to provide scientific and reasonable suggestions for passengers waiting and driver’s cruise scheduling,which can alleviate the relationship between taxis and passengers to a certain extent.supply-demand conflict.The main research contents of this paper are as follows:(1)Data preprocessing.According to the needs of the research content,the rules for data cleaning are established,effective trajectory data is extracted from the original taxi GPS data,and the drift points and deviation points in the trajectory data due to the influence of equipment accuracy are matched to the corresponding roads.To solve the problem of data error,the pickup and drop-off points in taxi operation are finally extracted from the trajectory data.(2)Analysis of travel time and space characteristics.The basic characteristics of taxi operation are explored from three perspectives: the temporal and spatial characteristics of travel,the temporal and spatial patterns,and the characteristics of external travel connections.In terms of time,the unit of day and hour is used to compare and analyze the operation law of taxis.In terms of space,the unit of Tyson regular hexagon grid is used to calculate and analyze the intensity of taxi trips and the law of attraction in different areas.Using the analysis methods of standard deviation ellipse and kernel density estimation,the connection law of taxis and external transportation hubs in Lanzhou City,as well as the spatiotemporal distribution characteristics of OD points,are studied.(3)Excavation of passenger waiting hot spots.The spatial-temporal sequence of passenger taxi hotspots is extracted by ST-DBSCAN spatiotemporal clustering algorithm,and the obtained spatial-temporal sequence is converted into a passenger taxi interest surface that can be given geographic spatial location attributes using K-nearest neighbor algorithm combined with the road network.The probability distribution model of passenger waiting time is established,the probability of passenger waiting time of each interest surface is calculated,and then the functional information entropy of the interest surface is calculated according to POI data.Finally,combined with the obtained indicators,the passengers’ interest surface for taxi rides is sorted by multiple factors,and then the hot spots for passengers waiting for the bus are obtained,which provides a reference for the subsequent identification of the driver’s passengercarrying area.(4)Identification of the driver’s passenger-carrying area.Combined with the spatial regular hexagonal grid,the taxi GPS data,POI data,and the passenger waiting hotspot area data obtained from the study are combined with multi-source data.The passenger area identification model,the fusion of multi-source data is used as the eigenvalue to calculate the attraction of taxis in the unit grid,and the parameter selection and model construction of the neural network are introduced in turn,and the calculation result is finally obtained.The results of this method are compared with the random forest model and the ridge regression model,and the Class A uncertainty and RMSE value under various conditions are calculated,which verifies the superiority and stability of the method proposed in this paper.