Research On Implementation Of Passenger Hotspots And Taxi Hotspots Based On Taxi Trajectory Data

At present,our country is vigorously advocating the development of public transport.Taxi because of its convenience has become the priority choice of daily travel for residents.However,on the one hand,the unbalanced spatial and temporal distribution of passenger sources leads to the high no-load rate of taxis and the difficulty of finding passengers.It reduces the operation efficiency of taxis and invisibly increases the operation cost of taxi drivers.On the other hand,for passengers who are not familiar with the urban traffic situation,there will be difficult to find the taxi.At present,there is no good application system for taxi drivers or passengers to solve these problems.A method is proposed in this thesis to increase the efficiency of taxi drivers and reduce the waiting time of passengers.It is based on Android platform.The main contents include:(1)Research on the algorithm of taxi trajectory residence point and passenger carrying point extraction based on speed and time.Firstly,the taxi GPS historical data are preprocessed.Then,the residence point and passenger carrying point are extracted according to the two factors of speed and time.Finally,the extracted residence points are clustered and optimized to obtain the core residence points.When extracting residence points,different clustering algorithms are used to optimize residence points for different patterns,which improves the quality of residence points and the time efficiency of next clustering.(2)Research on K-means clustering algorithm based on time division.By analyzing the trajectory data set of mobile objects published by Microsoft Asia Research Institute,we get the rule that people's activities change with time.The difference of attraction to people in different time periods and places leads to the difference of taxi demand.In order to provide drivers with high passenger carrying hot spots well and make passengers take taxis faster,this thesis carries out clustering analysis after dividing time periods.The experimental results show that the recommended points obtained by this method have high accuracy.(3)The prototype of the hot spot recommendation system for taxi drivers andpassengers based on Android is implemented.The mobile client of the system is developed based on Android platform.From the user's point of view,not only taxi drivers but also passengers are considered.So the app is divided into two versions-drive version and passage version.The main modules include information management module and hot spot acquisition module.