Research On Traffic Behavior And Taxi Travel Route Recommendation Based On GPS Data

In this paper,we mainly study recommending effective routes for vacant taxis by using GPS trajectory data.These routes enable them to carry passengers probably in a relatively short period of time.Urban traffic conditions are complicated and variable.The choice route of taxi driver has uncertainty and flexibility in different time and road segments.People's traveling patterns are also different in different periods and regions.Based on the above situation,the results of existing route recommendation methods are not satisfactory.The main work in this paper is as follows:We optimize "pick-up point" mining function.An ActualPickUpPoint algorithm is designed to mine actual pick-up point in potential clusters.ActualPickUpPoint algorithm gathers tight points with liner time complexity,and identity and filter noise data.Experiments on real taxi trajectory data verify the effectiveness and stability of the algorithm.We improve the way of traffic behavior mining.According to different passenger status,taxi trajectory is divided into two types.LCS based DBSCAN algorithm is used to mine different types of hotspot trajectory segments.Three main characteristics:passenger flow volume,distance and time are combined to comprehensively analyze traffic behavior on passenger seeking segments.Road travelling speed is utilized to analyze traffic behavior on passenger service segments.And then,the traffic behavior matrix of the whole road networks is constructed by using traffic behavior on different segments.In this paper,pick up point,hotspot trajectory cluster and traffic behavior are combined to design a weighted tree recommendation model:WTRM,which recommends reasonable routes for vacant taxis,enabling them to carry passengers with high probability.For a given query point,WTRM recommends an effective and complete route for vacant taxi among multiple possible routes.In order to enhance the rationality of route recommendation,WTRM add branch-reduction rules to remove edges that do not conform to rules and improve the computing efficiency.WTRM proposes multiple routes recommendation strategy,which can maximize the distance between recommended routes to recommend different routes for multiple empty taxis in the same region during the same period.The real GPS trajectory data is used to conduct experiments,which verify the effectiveness of WTRM.In this paper,we use the proposed model and algorithm to implement a pick-up point query,route recommendation and operation analysis system:EasyDriver.