Design And Implementation Of Taxi Service Application System Based On Hadoop Platform And Traiectory Mining

Urban traffic is the blood and lifeline of urban development,which is responsible for the migration of people and transportation of materials.With its convenience and flexibility,taxis become an important part of urban public transport.Taxis run on the roads,whose routes can be found in every corner of the city.The vehicle-mounted GPS devices records the spatial and temporal distribution of vehicles,which provide precious data value for urban road traffic system research.With the proposal of conception of urban computing,the researchers have found more and more hidden information in the trajectory data of the taxi,which plays an increasingly important role in the planning process of urban development.Meanwhile,the excavation of these data also has important guiding significance for the taxi service industry itself,which leads the new development direction for the industry.The annual amount of data generated by tens of thousands of taxis in the city is considerable.On the single machine,the algorithm and training parameters of the trajectory data must be low.At the same time,the huge amount of track data storage problems also restrict the progress of the next stage of research work.To this end,this paper is based on the passenger hotspot mining model based on the target event point,and combines the big data storage and computing platform to design and implement a taxi service application system.In this paper,the main work includes:(1)do research on data pretreatment technology,combined with large data processing platform for practice and design of trajectory data and passenger point extraction algorithm.(2)study clustering algorithm,design and implement parallel computing model with Spark platform.(3)of mass trajectory data using parallel pretreatment and cluster analysis,excavate the passenger hot spots and areas in the city,design and verify the route recommendation model based on passenger hot spots and areas.(4)based on the proposed model and data knowledge,the client software is developed to provide users with the most direct travel services.In order to verify the feasibility of this project,this paper deals with clustering effect intuitionistic visualization display and efficiency test.The parameter optimization and accuracy comparison test of the driving information recommendation service model are also carried out.The experimental results show that the optimized clustering algorithm greatly improve the efficiency of mass trajectory data processing and perform good clustering effect,and recommendation accuracy of the traffic information accuracy also remains in a relatively stable level.