Bicycle, pedestrian, and mixed-mode trail traffic

A performance assessment of demand models

Alireza Ermagun, Greg Lindsey, Tracy Hadden Loh

    Research output: Contribution to journalArticle

    4 Citations (Scopus)

    Abstract

    This study presents new trail demand models based on data collected between January 1, 2014 and February 16, 2016 at 32 locations in the seven major climatic regions in the continental U.S. We contribute fourfold to the literature on analysis of trail traffic demand. First, we develop a set of econometric models to predict average daily pedestrians (ADP), average daily bicyclists (ADB), and average daily mixed-mode traffic (ADM) using the 5 D's of the built environment (i.e., density, diversity, design, distance to transit, and destination accessibility), and socio-economic characteristics. Second, we test the performance of trail demand models in predicting ADB, ADP, and ADM using the leave-one-out cross-validation technique and compare the relative accuracy of the models. Third, we assess the performance of separate bicycle and pedestrian demand models in predicting mixed-mode travel demand. Fourth, we introduce a post-validation technique to advance the prediction accuracy of trail traffic demand models. The results indicate: (1) with only a few exceptions, ADP and ADB are correlated with different variables, and the magnitude of effects of variables that are the same varies significantly between the two modes; (2) The mean relative percentage error (MRPE) for bicyclist, pedestrian, and mixed-mode models equals 65.4%, 85.3%, and 45.9%; (3) Although using separate but integrated sensors to monitor bicycle and pedestrian traffic enables us to juxtapose the bicyclist demand with pedestrian demand, there is not a significant improvement in predicting total demand using these more expensive sensors; (4) A new post-validation procedure improved the demand models, reducing the MRPE of bicyclist, pedestrian, and mixed-mode models by 27.2%, 32.1%, and 14.1%. Overall, our models confirm that different variables are correlated with bicycle and pedestrian traffic volumes and that these modes need to be modeled separately. Our models can be used in practical applications such as selection of trail corridors and prioritization of investments where order-of-magnitude estimates suffice.

    Original languageEnglish (US)
    Pages (from-to)92-102
    Number of pages11
    JournalLandscape and Urban Planning
    Volume177
    DOIs
    StatePublished - Sep 1 2018

    Fingerprint

    performance assessment
    pedestrian
    bicycle
    traffic
    demand
    sensor
    travel demand
    climatic region
    prioritization
    econometrics
    accessibility

    Keywords

    • Average daily bicyclists
    • Average daily pedestrians
    • Built environment
    • Demand model
    • Mixed-mode
    • Trail traffic

    Cite this

    Bicycle, pedestrian, and mixed-mode trail traffic : A performance assessment of demand models. / Ermagun, Alireza; Lindsey, Greg; Hadden Loh, Tracy.

    In: Landscape and Urban Planning, Vol. 177, 01.09.2018, p. 92-102.

    Research output: Contribution to journalArticle

    @article{6485b0e094c248c6a2588d120d3e1f81,
    title = "Bicycle, pedestrian, and mixed-mode trail traffic: A performance assessment of demand models",
    abstract = "This study presents new trail demand models based on data collected between January 1, 2014 and February 16, 2016 at 32 locations in the seven major climatic regions in the continental U.S. We contribute fourfold to the literature on analysis of trail traffic demand. First, we develop a set of econometric models to predict average daily pedestrians (ADP), average daily bicyclists (ADB), and average daily mixed-mode traffic (ADM) using the 5 D's of the built environment (i.e., density, diversity, design, distance to transit, and destination accessibility), and socio-economic characteristics. Second, we test the performance of trail demand models in predicting ADB, ADP, and ADM using the leave-one-out cross-validation technique and compare the relative accuracy of the models. Third, we assess the performance of separate bicycle and pedestrian demand models in predicting mixed-mode travel demand. Fourth, we introduce a post-validation technique to advance the prediction accuracy of trail traffic demand models. The results indicate: (1) with only a few exceptions, ADP and ADB are correlated with different variables, and the magnitude of effects of variables that are the same varies significantly between the two modes; (2) The mean relative percentage error (MRPE) for bicyclist, pedestrian, and mixed-mode models equals 65.4{\%}, 85.3{\%}, and 45.9{\%}; (3) Although using separate but integrated sensors to monitor bicycle and pedestrian traffic enables us to juxtapose the bicyclist demand with pedestrian demand, there is not a significant improvement in predicting total demand using these more expensive sensors; (4) A new post-validation procedure improved the demand models, reducing the MRPE of bicyclist, pedestrian, and mixed-mode models by 27.2{\%}, 32.1{\%}, and 14.1{\%}. Overall, our models confirm that different variables are correlated with bicycle and pedestrian traffic volumes and that these modes need to be modeled separately. Our models can be used in practical applications such as selection of trail corridors and prioritization of investments where order-of-magnitude estimates suffice.",
    keywords = "Average daily bicyclists, Average daily pedestrians, Built environment, Demand model, Mixed-mode, Trail traffic",
    author = "Alireza Ermagun and Greg Lindsey and {Hadden Loh}, Tracy",
    year = "2018",
    month = "9",
    day = "1",
    doi = "10.1016/j.landurbplan.2018.05.006",
    language = "English (US)",
    volume = "177",
    pages = "92--102",
    journal = "Landscape and Urban Planning",
    issn = "0169-2046",
    publisher = "Elsevier",

    }

    TY - JOUR

    T1 - Bicycle, pedestrian, and mixed-mode trail traffic

    T2 - A performance assessment of demand models

    AU - Ermagun, Alireza

    AU - Lindsey, Greg

    AU - Hadden Loh, Tracy

    PY - 2018/9/1

    Y1 - 2018/9/1

    N2 - This study presents new trail demand models based on data collected between January 1, 2014 and February 16, 2016 at 32 locations in the seven major climatic regions in the continental U.S. We contribute fourfold to the literature on analysis of trail traffic demand. First, we develop a set of econometric models to predict average daily pedestrians (ADP), average daily bicyclists (ADB), and average daily mixed-mode traffic (ADM) using the 5 D's of the built environment (i.e., density, diversity, design, distance to transit, and destination accessibility), and socio-economic characteristics. Second, we test the performance of trail demand models in predicting ADB, ADP, and ADM using the leave-one-out cross-validation technique and compare the relative accuracy of the models. Third, we assess the performance of separate bicycle and pedestrian demand models in predicting mixed-mode travel demand. Fourth, we introduce a post-validation technique to advance the prediction accuracy of trail traffic demand models. The results indicate: (1) with only a few exceptions, ADP and ADB are correlated with different variables, and the magnitude of effects of variables that are the same varies significantly between the two modes; (2) The mean relative percentage error (MRPE) for bicyclist, pedestrian, and mixed-mode models equals 65.4%, 85.3%, and 45.9%; (3) Although using separate but integrated sensors to monitor bicycle and pedestrian traffic enables us to juxtapose the bicyclist demand with pedestrian demand, there is not a significant improvement in predicting total demand using these more expensive sensors; (4) A new post-validation procedure improved the demand models, reducing the MRPE of bicyclist, pedestrian, and mixed-mode models by 27.2%, 32.1%, and 14.1%. Overall, our models confirm that different variables are correlated with bicycle and pedestrian traffic volumes and that these modes need to be modeled separately. Our models can be used in practical applications such as selection of trail corridors and prioritization of investments where order-of-magnitude estimates suffice.

    AB - This study presents new trail demand models based on data collected between January 1, 2014 and February 16, 2016 at 32 locations in the seven major climatic regions in the continental U.S. We contribute fourfold to the literature on analysis of trail traffic demand. First, we develop a set of econometric models to predict average daily pedestrians (ADP), average daily bicyclists (ADB), and average daily mixed-mode traffic (ADM) using the 5 D's of the built environment (i.e., density, diversity, design, distance to transit, and destination accessibility), and socio-economic characteristics. Second, we test the performance of trail demand models in predicting ADB, ADP, and ADM using the leave-one-out cross-validation technique and compare the relative accuracy of the models. Third, we assess the performance of separate bicycle and pedestrian demand models in predicting mixed-mode travel demand. Fourth, we introduce a post-validation technique to advance the prediction accuracy of trail traffic demand models. The results indicate: (1) with only a few exceptions, ADP and ADB are correlated with different variables, and the magnitude of effects of variables that are the same varies significantly between the two modes; (2) The mean relative percentage error (MRPE) for bicyclist, pedestrian, and mixed-mode models equals 65.4%, 85.3%, and 45.9%; (3) Although using separate but integrated sensors to monitor bicycle and pedestrian traffic enables us to juxtapose the bicyclist demand with pedestrian demand, there is not a significant improvement in predicting total demand using these more expensive sensors; (4) A new post-validation procedure improved the demand models, reducing the MRPE of bicyclist, pedestrian, and mixed-mode models by 27.2%, 32.1%, and 14.1%. Overall, our models confirm that different variables are correlated with bicycle and pedestrian traffic volumes and that these modes need to be modeled separately. Our models can be used in practical applications such as selection of trail corridors and prioritization of investments where order-of-magnitude estimates suffice.

    KW - Average daily bicyclists

    KW - Average daily pedestrians

    KW - Built environment

    KW - Demand model

    KW - Mixed-mode

    KW - Trail traffic

    UR - http://www.scopus.com/inward/record.url?scp=85047064795&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=85047064795&partnerID=8YFLogxK

    U2 - 10.1016/j.landurbplan.2018.05.006

    DO - 10.1016/j.landurbplan.2018.05.006

    M3 - Article

    VL - 177

    SP - 92

    EP - 102

    JO - Landscape and Urban Planning

    JF - Landscape and Urban Planning

    SN - 0169-2046

    ER -