| Bus passenger flow data is the base data of operating and scheduling bus for bus passenger transportation planning department. Therefore, accurate grasp for the bus passenger flow and predicting the future bus passenger flow is very important. But the method for the bus passenger flow obtaining is not comprehensive and timeliness. In this paper, we study how to analysis and predict the bus passenger flow using Hadoop and bus IC data, GPS data and weather data to improve the computational efficiency of the model prediction under the premise of guarantee prediction accuracy.Firstly, based on the IC card data and GPS data we analysis the bus passenger information of getting on the bus,then based on the passenger travel chain we analysis the bus passenger information of getting off the bus. In the analysis of getting on the bus, based on time matching of bus IC data and bus GPS data to determine the location of getting on the bus. In the analysis of getting off the bus, based on passenger travel chain to judge the station of getting off the bus. Secondly, this paper analyzes the impact factors of bus passenger flow including periodicity, holidays and weather. Finally, based on the above analysis, this paper predict the bus passenger flow in the future. Comparating all kinds of predicting model to each other, the paper selected BP artificial neural network model to predict for the ability of nonlinear mapping and learning bility. The paper consider the historical data, temperature and weather factors.Against disadvantages of the time-consuming problem of BP artificial neural network training, the paper use the Map Reduce model to implement the algorithm. Experimental results show that BP artificial neural network based on Map Reduce implementation to predict future bus passenger flow is a effective algorithm. Compared to other algorithms, the algorithm.Compared improve the efficiency of the algorithm operation and Guaranteeing the accuracy of forecast. |
PDF Full Text Request