A precise prediction of transportation time is important to help both passengers to plan their trips and bus operations control to make an effective fleet management. In this study, we make use of GPS data from a public transportation bus line to develop a public bus arrival time prediction at any distance along the route. With large and complex information, deep neural network model (DNN) is used to get high prediction accuracy. In this paper, variables and structures of the proposed DNN model are presented. The performance of the proposed model is evaluated by conducting real BMTA-8 bus data in Bangkok, Thailand, and comparing the result with the currently used ordinary least square (OLS) regression model. The result shows that the proposed DNN model is more accurate than the OLS regression model around 55% for mean absolute percentage error (MAPE). It outperforms the current prediction method of the studied bus line, and it is feasible and applicable for bus travel time prediction of any route.


    Access

    Check access

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    Bus arrival time prediction at any distance of bus route using deep neural network model




    Publication date :

    2017-10-01


    Size :

    1110537 byte





    Type of media :

    Conference paper


    Type of material :

    Electronic Resource


    Language :

    English




    Bus arrival time prediction using mixed multi-route arrival time data at previous stop

    Xuedong Hua / Wei Wang / Yinhai Wang et al. | DOAJ | 2018

    Free access



    Real-time bus arrival prediction method based on route tracks

    LI GANG / ZHOU HUIBO / LIU YUE | European Patent Office | 2021

    Free access