Hot Routes Mining And Passenger Routes Recommendation Based On Taxi GPS Trajectory

Taxi passenger trajectory directly reflects the driving state of the vehicle and the travel law of residents.Hot routes mining not only to provides guidance for the taxi passengers,effectively increase the revenue of taxi drivers,but also provides guidance on the routes that taxis take after picking up passengers.It can relieve urban traffic congestion.It is of great value for traffic management,planning,residents' behavior pattern discovery,and taxi driver recommendation to excavate the hot routes.This paper takes the GPS trajectory data of 3000 taxis in Lanzhou city as the research object.According to the taxi passenger trajectory,a novel algorithm based on spatial-temporal similarity clustering is proposed to mine the hot routes.The distribution characteristics of hot routes in Lanzhou are studied.Combining the shortest path between passenger points and urban traffic congestion index,the algorithm of taxi carrying route is proposed to recommend the reasonable route for taxi drivers.The main researchs are summarized as follows:(1)The kernel passenger trajectory algorithm and the trajectory spatial-temporal similarity algorithm are proposed.Preprocessing the original data of GPS trajectory and extracting passenger trajectories from GPS trajectory data,an algorithm based on reducing complexity to extract the kernel passenger trajectory is proposed combining the original network topology.A trajectory spatial-temporal similarity algorithm,which can reflect both spatial and temporal attributes of passenger travel,is proposed.(2)A mining algorithm for hot routes based on spatial-temporal similarity clustering is proposed.According to the proposed similarity measure algorithm,the spatial similarity,temporal similarity and spatial-temporal similarity of kernel trajectories are calculated.We use the spectral clustering algorithm to cluster the passenger trajectory.According to the clustering results,the spatial distribution of the hot passenger path is obtained,and the differences between weekdays and weekends are analyzed.The experimental results show that the proposed mining algorithm can effectively and quickly find the distribution of hot routes,and can provide important decision support for urban traffic management and road network adjustment.(3)A taxi passenger routes algorithm based on the shortest path and urban traffic congestion index is proposed.According to the GPS trajectory,the pick-up points are extracted,and the k-mediod clustering algorithm is used to obtain the passenger points.According to the distance between taxi passenger points and congestion index,the optimal passenger path combining with Dijkstra algorithm,is recommended and analyzed visually.The experimental results show that the proposed mining algorithm can effectively and reasonably recommend the passenger path for taxi drivers.