Comprehensive approach for the sensitivity analysis of high-dimensional and computationally expensive traffic simulation models

Qiao Ge, Biagio Ciuffo, Monica Menendez

    Research output: Contribution to journalArticle

    Abstract

    The reliability of traffic model results is strictly connected to the quality of its calibration. A challenge arising in this context concerns the selection of the most influential input parameters. A model sensitivity analysis should be used with this aim. However, because of the limitations of time and computational resources, a proper sensitivity analysis is rarely performed in common practice. A recent study introduced a methodology based on Gaussian process metamodels for the sensitivity analysis of computationally expensive traffic simulation models. The main limitation was a dependence on model dimensionality. When the model has more than about 15 to 20 parameters, estimation of a Gaussian process metamodel (also known as a Kriging metamodel) may become problematic. In this paper, the Kriging-based approach is coupled with a recently developed approach, quasi-optimized trajectorybased elementary effects (quasi-OTEE), for the sensitivity analysis of computationally expensive models. The quasi-OTEE sensitivity analysis can be used to identify the whole subset of sensitive parameters of a high-dimensional model, and the Kriging-based sensitivity analysis can then be used to refine the analysis and to rank the different parameters of the subset in a more reliable way. Application of this new sequential sensitivity analysis method is illustrated with the Wiedemann-74 carfollowing model. Results show that the new method requires 40 times fewer model evaluations than a standard variance-based sensitivity analysis to identify the influential parameters and their ranks.

    Original languageEnglish (US)
    Pages (from-to)121-130
    Number of pages10
    JournalTransportation Research Record
    Volume2422
    DOIs
    StatePublished - Jan 1 2014

    Fingerprint

    Sensitivity analysis
    Parameter estimation
    Calibration

    ASJC Scopus subject areas

    • Civil and Structural Engineering
    • Mechanical Engineering

    Cite this

    Comprehensive approach for the sensitivity analysis of high-dimensional and computationally expensive traffic simulation models. / Ge, Qiao; Ciuffo, Biagio; Menendez, Monica.

    In: Transportation Research Record, Vol. 2422, 01.01.2014, p. 121-130.

    Research output: Contribution to journalArticle

    @article{8b8a1fed3ddc480cab113da94f4b7f2a,
    title = "Comprehensive approach for the sensitivity analysis of high-dimensional and computationally expensive traffic simulation models",
    abstract = "The reliability of traffic model results is strictly connected to the quality of its calibration. A challenge arising in this context concerns the selection of the most influential input parameters. A model sensitivity analysis should be used with this aim. However, because of the limitations of time and computational resources, a proper sensitivity analysis is rarely performed in common practice. A recent study introduced a methodology based on Gaussian process metamodels for the sensitivity analysis of computationally expensive traffic simulation models. The main limitation was a dependence on model dimensionality. When the model has more than about 15 to 20 parameters, estimation of a Gaussian process metamodel (also known as a Kriging metamodel) may become problematic. In this paper, the Kriging-based approach is coupled with a recently developed approach, quasi-optimized trajectorybased elementary effects (quasi-OTEE), for the sensitivity analysis of computationally expensive models. The quasi-OTEE sensitivity analysis can be used to identify the whole subset of sensitive parameters of a high-dimensional model, and the Kriging-based sensitivity analysis can then be used to refine the analysis and to rank the different parameters of the subset in a more reliable way. Application of this new sequential sensitivity analysis method is illustrated with the Wiedemann-74 carfollowing model. Results show that the new method requires 40 times fewer model evaluations than a standard variance-based sensitivity analysis to identify the influential parameters and their ranks.",
    author = "Qiao Ge and Biagio Ciuffo and Monica Menendez",
    year = "2014",
    month = "1",
    day = "1",
    doi = "10.3141/2422-14",
    language = "English (US)",
    volume = "2422",
    pages = "121--130",
    journal = "Transportation Research Record",
    issn = "0361-1981",
    publisher = "US National Research Council",

    }

    TY - JOUR

    T1 - Comprehensive approach for the sensitivity analysis of high-dimensional and computationally expensive traffic simulation models

    AU - Ge, Qiao

    AU - Ciuffo, Biagio

    AU - Menendez, Monica

    PY - 2014/1/1

    Y1 - 2014/1/1

    N2 - The reliability of traffic model results is strictly connected to the quality of its calibration. A challenge arising in this context concerns the selection of the most influential input parameters. A model sensitivity analysis should be used with this aim. However, because of the limitations of time and computational resources, a proper sensitivity analysis is rarely performed in common practice. A recent study introduced a methodology based on Gaussian process metamodels for the sensitivity analysis of computationally expensive traffic simulation models. The main limitation was a dependence on model dimensionality. When the model has more than about 15 to 20 parameters, estimation of a Gaussian process metamodel (also known as a Kriging metamodel) may become problematic. In this paper, the Kriging-based approach is coupled with a recently developed approach, quasi-optimized trajectorybased elementary effects (quasi-OTEE), for the sensitivity analysis of computationally expensive models. The quasi-OTEE sensitivity analysis can be used to identify the whole subset of sensitive parameters of a high-dimensional model, and the Kriging-based sensitivity analysis can then be used to refine the analysis and to rank the different parameters of the subset in a more reliable way. Application of this new sequential sensitivity analysis method is illustrated with the Wiedemann-74 carfollowing model. Results show that the new method requires 40 times fewer model evaluations than a standard variance-based sensitivity analysis to identify the influential parameters and their ranks.

    AB - The reliability of traffic model results is strictly connected to the quality of its calibration. A challenge arising in this context concerns the selection of the most influential input parameters. A model sensitivity analysis should be used with this aim. However, because of the limitations of time and computational resources, a proper sensitivity analysis is rarely performed in common practice. A recent study introduced a methodology based on Gaussian process metamodels for the sensitivity analysis of computationally expensive traffic simulation models. The main limitation was a dependence on model dimensionality. When the model has more than about 15 to 20 parameters, estimation of a Gaussian process metamodel (also known as a Kriging metamodel) may become problematic. In this paper, the Kriging-based approach is coupled with a recently developed approach, quasi-optimized trajectorybased elementary effects (quasi-OTEE), for the sensitivity analysis of computationally expensive models. The quasi-OTEE sensitivity analysis can be used to identify the whole subset of sensitive parameters of a high-dimensional model, and the Kriging-based sensitivity analysis can then be used to refine the analysis and to rank the different parameters of the subset in a more reliable way. Application of this new sequential sensitivity analysis method is illustrated with the Wiedemann-74 carfollowing model. Results show that the new method requires 40 times fewer model evaluations than a standard variance-based sensitivity analysis to identify the influential parameters and their ranks.

    UR - http://www.scopus.com/inward/record.url?scp=84979082331&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=84979082331&partnerID=8YFLogxK

    U2 - 10.3141/2422-14

    DO - 10.3141/2422-14

    M3 - Article

    VL - 2422

    SP - 121

    EP - 130

    JO - Transportation Research Record

    JF - Transportation Research Record

    SN - 0361-1981

    ER -