Short-term traffic speed prediction plays a crucial role in intelligent transportation systems, e.g., vehicle navigation. Short-term traffic speed forecasting used to be analyzed only considering temporal factors. In this paper, we take both temporal and spatial factors into account. Features of the upstream and downstream locations are considered to establish the random forest (RF) and support vector regression (SVR) models. In order to improve the prediction accuracy, two brand new variables (i.e., absolute congestion and relative congestion) calculated from raw data are proposed. The RF is built to test the new variables’ influence on the prediction accuracy. Traffic data extracted from the California Freeway Performance Measurement System (PeMS) is used to build and test the speed prediction models. It is verified that both upstream and downstream speed variables are able to improve model accuracy, and the two proposed new features are helpful to further improving the accuracy.


    Access

    Check access

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    Random Forests for Freeway Short-Term Traffic Speed Prediction


    Contributors:

    Conference:

    17th COTA International Conference of Transportation Professionals ; 2017 ; Shanghai, China


    Published in:

    CICTP 2017 ; 120-130


    Publication date :

    2018-01-18




    Type of media :

    Conference paper


    Type of material :

    Electronic Resource


    Language :

    English