Urban Public Traffic Flow Analysis And Application Research Based On Data Mining

In recent years,with China's sustained economic development,urbanization continues to accelerate,the city is also expanding the scale,there is a sharp increase in personal vehicle ownership and road traffic,which leads to the increasingly prominent contradiction between traffic supply and traffic demand.However,an important method to solve these urban traffic problems is a priority development of public transport,and only comprehensive,accurate and timely understanding of passenger flow characteristics in the process of public transportation operation,can help the public transit managers to make more scientific public transportation planning and operation decision-making,so as to truly improve the traffic condition.The thesis selects the research data from public transport passenger traffic data of Shenzhen City,the research method is combined with the advanced data mining method and the traditional statistical methods,so it can get the law of public transport passenger flow and passenger flow indicators of public transport operators,and realize the public transport passenger flow forecast and early warning function by the data mining of massive IC card data.The thesis presents the analysis framework to complish the research of public transport passenger flow data based on data mining,First,it collects the corresponding research data,such as IC card data,GPS data of bus and other relative data to build the database,and clean the data to improve the efficiency of data mining;secondly,according to the research purpose,it uses the cluster analysis to establish the on-station matching model,the off-station predicting model and the run-times counting model;and then combine with the statistical analysis to get the public transport passenger flow and passenger travel law;finally,based on the preliminary data mining work and the public transportation passenger flow and passenger flow law,this paper establishs three different passenger flow prediction models(the prediction model includes artificial neural network,support vector machine,decision tree)to finish the depth mining research,then compare the errors to select the optimal prediction model,so based onthe passenger flow forecast data and historical traffic data,this paper establishs the model of passenger flow threshold to realize the real time warning of public traffic.