Medium-term public transit route ridership forecasting: What, how and why? A case study in Lyon

Oscar Egu and Patrick Bonnel
Additional contact information
Oscar Egu: LAET - Laboratoire Aménagement Économie Transports - UL2 - Université Lumière - Lyon 2 - ENTPE - École Nationale des Travaux Publics de l'État - CNRS - Centre National de la Recherche Scientifique

Post-Print from HAL

Abstract: Demand forecasting is an essential task in many industries and the transportation sector is no exception. This is because accurate forecasts are a fundamental aspect of any rationale planning process and an essential component of intelligent transportation systems. In the context of public transit, forecasts are needed to support different level of planning and organisational processes. Short-term forecast, typically a few hours in the future, are developed to support real-time operations. Long-term forecast, typically 5 years or more in the future, are essential for strategic planning. Those two forecast horizons have been widely studied by the academic community but surprisingly little research deal with forecast between those two ranges. The objective of this paper is therefore twofold. First, we proposed a generic modelling approach to forecast next 365 days ridership in a public transit network at different levels of spatiotemporal aggregation. Second, we illustrate how such models can assist public transit operators and transit agencies in monitoring ridership and supporting recurrent tactical planning tasks. The proposed formulation is based on a multiplicative decomposition that combines tree-based models with trend forecasting. The evaluation of models on unseen data proves that this approach generates coherent forecast. Different use cases are then depicted. They demonstrate that the resulting forecast can support various recurrent tactical tasks such as setting future goals, monitoring ridership or supporting the definition of service provision. Overall, this study contributes to the growing literature on the use of automated data collection. It confirms that more sophisticated statistical methods can help to improve public transportation planning and enhance data-driven decision making.

Keywords: Public transit; Ridership forecasting; Machine learning; Smart card data; Transport planning (search for similar items in EconPapers)
Date: 2021-05
Note: View the original document on HAL open archive server: https://shs.hal.science/halshs-04233578
References: View references in EconPapers View complete reference list from CitEc
Citations:

Published in Transport Policy, 2021, 105, pp.124-133. ⟨10.1016/j.tranpol.2021.03.002⟩

Downloads: (external link)
https://shs.hal.science/halshs-04233578/document (application/pdf)

Related works:
Journal Article: Medium-term public transit route ridership forecasting: What, how and why? A case study in Lyon (2021)
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:hal:journl:halshs-04233578

DOI: 10.1016/j.tranpol.2021.03.002

Access Statistics for this paper

More papers in Post-Print from HAL
Bibliographic data for series maintained by CCSD (hal@ccsd.cnrs.fr).