Feature Mining Of Travel Route Sets Based On Taxi GPS Data

At present,more and more taxis in China are equipped with GPS.The GPS data contains important information such as the law of vehicle driving,and mining the hidden information in the data has important guiding significance for vehicle travel.In recent years,most of the research on taxi GPS data has focused on the research of route selection behavior modeling and route set generation algorithms,and few scholars have studied the characteristics of travel route sets.Based on this,this paper studies the characteristics of the travel path sets in order to mine the operating rules implied in the taxi GPS data,to explore the operating conditions of urban roads,and to further extract the experience and wisdom of taxi drivers.In the face of congested traffic flow in the city,based on the experience of taxi drivers and knowledge of the road network structure,the best driving route to shorten the congestion time can be found as much as possible,which provides good suggestions for solving urban traffic congestion problems.The research results are as follows:First of all,this paper takes the GPS data of taxis in Xi’an city as the research object,cleans and processes the abnormal GPS data of the status of taxi passengers,extracts the up and down passenger points of the taxi,selects DBSCAN algorithm to determine the hot spots of taxi trips,and uses Python software to extract OD travel path in hot spots.Then,the distribution range of the travel path is determined,and representative OD pairs are selected from the hotspot area,and the path distribution trend between the OD pairs is analyzed.Using the standard deviation ellipse fitting results,it is found that the travel path between the OD pairs belongs to the ellipse distribution,and the ellipse search algorithm is used to limit the path search area,so that the path search between OD pairs are limited to a specific range.Finally,the characteristic indexes of seven travel path sets are selected,including the straight-line distance of the path,the number of intersections,the number of turns,the level of the path,the time of the path,the time of departure,and the circuitous degree of the path.Through the CART algorithm,the threshold of the characteristic index is determined,and the maximum number of left turns per kilometer,the maximum time consumption per kilometer,and the maximum number of path intersections per kilometer that the driver can accept under different OD travel distances are found,and analyzes the importance of the characteristic index considered by the driver when choosing the route.