modeling

Ensemble Models of For-Hire Vehicle Trips

| David Levinson

Recently published:

Wu, Hao and Levinson, David (2022) Ensemble Models of For-Hire Vehicle Trips. Frontiers in Future Transportation. 3 [DOI]

Ensemble forecasting is class of modeling approaches that combines different data sources, models of different types, with different assumptions, and/or pattern recognition methods. By comprehensively pooling information from multiple sources, analyzed with different techniques, ensemble models can be more accurate, and can better account for different sources of real-world uncertainties. The share of for-hire vehicle (FHV) trips increased rapidly in recent years. This paper applies ensemble models to predicting for-hire vehicle (FHV) trips in Chicago and New York City, showing that properly applied ensemble models can improve forecast accuracy beyond the best single model.

The Ensemble Approach to Forecasting: A Review and Synthesis

| David Levinson

Recently published:

  • Wu, Hao, and Levinson, D. (2021) The Ensemble Approach to Forecasting: A Review and Synthesis. Transportation Research part C. Volume 132, 103357 [doi] [Author Link]

    • Highlights

    • Review and synthesize methods of ensemble forecasting with a unifying framework.
    • As decision support tools, ensemble models systematically account for uncertainties.
    • Ensemble methods can include combining models, data, and ensemble of ensembles.
    • Transport ensemble models have the potential for improving accuracy and reliability.

    Abstract

    Ensemble forecasting is a modeling approach that combines data sources, models of different types, with alternative assumptions, using distinct pattern recognition methods. The aim is to use all available information in predictions, without the limiting and arbitrary choices and dependencies resulting from a single statistical or machine learning approach or a single functional form, or results from a limited data source. Uncertainties are systematically accounted for. Outputs of ensemble models can be presented as a range of possibilities, to indicate the amount of uncertainty in modeling. We review methods and applications of ensemble models both within and outside of transport research. The review finds that ensemble forecasting generally improves forecast accuracy, robustness in many fields, particularly in weather forecasting where the method originated. We note that ensemble methods are highly siloed across different disciplines, and both the knowledge and application of ensemble forecasting are lacking in transport. In this paper we review and synthesize methods of ensemble forecasting with a unifying framework, categorizing ensemble methods into two broad and not mutually exclusive categories, namely combining models, and combining data; this framework further extends to ensembles of ensembles. We apply ensemble forecasting to transport related cases, which shows the potential of ensemble models in improving forecast accuracy and reliability. This paper sheds light on the apparatus of ensemble forecasting, which we hope contributes to the better understanding and wider adoption of ensemble models.

    Fig. 1. Methods of combining data and models.

    Induced model complexity

    | David Levinson

    The End of Traffic and the Future of Access: A Roadmap to the New Transport Landscape. By David M. Levinson and Kevin J. Krizek.
    The End of Traffic and the Future of Access: A Roadmap to the New Transport Landscape. By David M. Levinson and Kevin J. Krizek.

    When I was a naive young modeler, running the Travel and Travel/2 models for the Montgomery County Planning Departments, regional travel demand models took up to 24 hours to run in full form. Talking with modelers today, it seems models still take on the order of 24 hours to run. Why?

    I posit “Induced Complexity.” When we build a road, we induce demand, travelers who were previously priced off the road due to congestion or extra travel time now switch times of day, routes, modes, and destinations to take advantage of the capacity, and new development is pursued. Similarly, when we get a bigger computer, we can either use it to run the same models faster, or to run more complicated models. It seems the profession leans to the latter. The complexity is in terms of the number of Transportation Analysis Zones, or in the number of Times of Day, or in the number of model components that are considered, or the degree of precision required in equilibrium.

    This induced complexity is real, and like induced demand is not necessarily a bad thing (if the complexity improves accuracy, it is a good thing), but it is a thing we should all be cognizant of.

    Journal of Transport and Land Use.

    | David Levinson

    We are pleased to announce the Journal of Transport and Land Use.
    What, you ask? Another journal amidst an already overcrowded field?
    Yes, we respond enthusiastically! Several journals touch on the interaction of transport and land use; however, they do so peripherally. This new venue puts both transport and land use front and center. We seek to be the leading outlet for research at the interdisciplinary intersection of these two domains, including work from the domains of engineering, planning, modeling, behavior, economics, geography, regional science, sociology, architecture and design, network science, and complex systems.
    The Journal of Transport and Land Use (JTLU) will be peer-reviewed, web-based, open-content, subscription-free, and free to contribute. All of this is enabled by support from the Center for Transportation Studies at the University of Minnesota, where the journal will be housed. The advantages of this new journal and new process are several:
    1. With a rigorous peer-review process, only quality papers that meet scientific standards will be published within the journal.
    2. By being web-based (and web-only), we reduce costs significantly compared with paper journals. Web-based publication allows a much faster turnaround time than paper publication. Our goal is six weeks between submission and first reviews returned to the author. Being web-based also allows the inclusion of full color graphics and multi-media content, and the inclusion of datasets with the publication.
    3. By being open-content, papers published in JTLU can be freely distributed (with attribution), increasing the value of papers published in the journal, and increasing their likelihood of being used in course readers and being read by the public.
    4. By being subscription-free, we overcome a fundamental problem of today’s expensive journals published by for-profit publishers, which many libraries can no longer subscribe to.
    5. By being free-to-contribute, we overcome the burden of the open-content journals that charge the authors to publish their paper.
    We are now soliciting papers covering topics at the intersection of transport and land use. Details about the journal, its editorial process, and paper submission can be found at the journal’s website http://www.jtlu.org .
    If you are interested in organizing a special issue, please contact one of the editors.
    There will be a meeting at the World Conference on Transport Research in Berkeley to discuss the journal, contact the editors for details.
    We look forward to any comments, questions, or suggestions you may have.
    Sincerely,
    David Levinson and Kevin Krizek
    David Levinson
    Richard P. Braun/CTS Chair in Transportation Engineering
    Director Networks, Economics, and Urban Systems (Nexus) Research Group
    University of Minnesota (612) 625-6354
    dlevinson@umn.edu
    http://nexus.umn.edu
    Kevin J. Krizek
    Associate Professor, Urban Planning & Civil Engineering
    University of Minnesota (612) 625 – 7318
    http://www.kevinjkrizek.org