Research On Urban Residents' Travel Time And Space Feature Mining Based On Taxi GPS Data

With the rapid development of GPS positioning,satellite navigation,wireless communication technology and so on,civil GPS and other positioning devices are becoming more and more popular and widely used.These GPS positioning devices as well as location-based services(LBS)for a variety of applications generate a large amount of trajectory data from moving objects.With the development of spatial data mining technology and geographic information technology,it is possible to study the trajectory data of a large number of moving objects.At the same time,the rapid development of the city,urban space in the expansion,the development of urban traffic has gradually become a judge of the scale of urban development and potential.Nowadays,the crux of urban traffic congestion problem is often concentrated in the hot spots of the city,how to identify the residents of the hot areas of travel and the study of the spatial and temporal characteristics of the residents travel behavior has been paid more and more attention.However,these studies often use the popular travel survey questionnaire statistical research,there are many limitations and one sidedness,and moving object trajectory data acquisition is more and more fast,storage more convenient,which makes the research on city traffic and travel behavior has entered a peak period.Methods using spatio-temporal data mining research on taxi GPS trajectory data,explore the law of the travel behaviors implied,solve the road congestion and other traffic planning problems,and for location based services for the relevant departments to provide scientific reference.In this paper,data mining research on taxi GPS trajectory,according to two working days and rest days,will rule all day long division of taxi operation in different period characteristics,also found that hot area residents travel in space,and the passenger taxi trajectory clustering characteristics of passenger travel mode for mobile.Therefore,this paper finds:(1)For taxi trajectory data pretreatment,time characteristics from sunrise for quantity,long distance,passenger,passenger loading rate,combining the four aspects of working days and rest days two levels to conclude that the taxi travel:sooner or later work day travel peak characteristics significantly;the rest of the day peak travel relatively lag,and travel demand is relatively stable.Passenger length is mainly concentrated in 5-20 minutes,the passenger distance between 3-15Km,and the main day of travel for short distance travel,the rest of the trip is mainly to travel halfway.Work on no-load rate fluctuation characteristics,the travel law is consistent with the working day;the rest day of the no-load rate relatively stable feature,in the morning and evening peak stage,working day load rate were lower than the rest day,afternoon and evening weekday load rate is higher than the rest of the day.(2)Combined with spatial point pattern analysis method and spatial autocorrelation analysis technology,this paper carries out the analysis of the passenger hot spots in the city.Through the global Moran I index,s' Moran 's I plot to analyze the concentration of taxi passengers in the space,"the global Moran week s I index s I index,found that low Moran than the rest of the day",that the rest of the day the taxi passengers scattered.To detect the city taxi hot regional distribution under the passenger through kernel density estimation method,the results showed that the time of day on and off hot area of the main traffic hub train station area and the commercial center of Xinjiekou region,the other on and off focus is mainly concentrated in the tower,the Confucius temple,the palace culture area and the rest of the day on and off with hot working days compared to disperse.(3)In the traditional OPTICS(Ordering points to identify the clustering structure)algorithm based on trajectory data is proposed according to the characteristics of suitable for mass trajectory space clustering TR-OPTICS algorithm(Trajectory OPTICS).The selection method of passenger taxi trajectory trajectory as the research object,through trajectory feature selection using MDL(Minimum Description Length)two division of track,bycalculating the sub trajectory of horizontal distance between the vertical distance,angle and distance to measure the similarity of the trajectory.In the clustering algorithm,the outer rectangle is used as the search space of the core trajectory,and the distance between the core and the path of the trajectory is redefined.And after many experiments with the traditional OPTICS algorithm,DBSCAN(Density-Based Spatial Clustering of Applications with Noise)algorithm comparison,the proposed TR-OPTICS algorithm in the efficiency of the algorithm is better than the other two algorithms in clustering results,this algorithm can effectively find the passenger sub track clusters and the clustering result is better than the other two algorithms.