Missing values are an important issue for traffic data programs and lead to difficulties in data analyses and application. Previous research accurately imputed missing hourly volumes of a range from a few days up to one and a half weeks for various types of traffic counts. However, imputation errors of the applied hourly models increase dramatically after some reliable imputing periods; this indicates their limitations in real applications. Therefore, this study investigates pattern-matching algorithms to fill large missing data intervals. The algorithms use historical data to develop a series of candidate patterns and compare them to patterns with missing data. Data from the best-fit pattern is then used to estimate the missing data. The algorithms show improved accuracy over traditional models used in practice and prove suitable to impute missing hourly, daily, and monthly traffic volume data.


    Zugriff

    Download

    Verfügbarkeit in meiner Bibliothek prüfen

    Bestellung bei Subito €


    Exportieren, teilen und zitieren



    Titel :

    Matching Hourly, Daily, and Monthly Traffic Patterns to Estimate Missing Volume Data


    Weitere Titelangaben:

    Transportation Research Record


    Beteiligte:
    Zhong, Ming (Autor:in) / Sharma, Satish (Autor:in)


    Erscheinungsdatum :

    2006-01-01




    Medientyp :

    Aufsatz (Zeitschrift)


    Format :

    Elektronische Ressource


    Sprache :

    Englisch