Transit service planning has traditionally focused on peak trips and the needs of “rush hour” commuters rather than off-peak travel. Often, off-peak trips are taken by shift-based essential workers and those who cannot or do not drive. The COVID-19 pandemic further underscored the need for a closer examination of these trips to improve social equity.
To better understand transit users who ride during non-peak times, researchers with the Transit Impacts Research Program contrasted peak with non-peak travel behaviors in Minnesota’s Twin Cities metropolitan area. The three-phase study, “Transfer Behavior and Off-Peak Commutes,” provides new insights into evolving transit behaviors and highlights the importance of the transitway system in facilitating efficient travel.
“By examining the behavior of riders during off-peak hours and how transitways support these journeys, we can develop a more comprehensive understanding of underrepresented travel patterns to help adapt the transit system to increase transportation equity,” says lead researcher Alireza Khani, associate professor of civil, environmental, and geo- engineering and a CTS scholar.
A new research brief summarizes the report’s key findings and recommendations. In phase one, researchers used riders’ data to compare peak and off-peak trip behavior. Among the findings: peak-time riders traveled a longer average distance than off-peak transit riders and, especially during the morning peak, took home- and work-based trips. Phase two focused on trips on the METRO system of light-rail and bus rapid transitways; researchers found that travel time and transfers drove decisions to choose transitways over other routes. Phase three leveraged machine-learning techniques to examine longer-term patterns spanning the pandemic from 2018-2023.
The study’s recommendations highlight that it’s critical that transit agencies continuously monitor riders’ short-term and long-term behaviors to ensure that services are matched to evolving community needs. Researchers also suggest that the machine-learning models developed in this study could greatly assist with this process.
—John Siqveland, CTS communications director
