.svg)
Mobility Data for Transportation Research
Mobility data records where a device was at a given moment, which researchers reassemble into trips: origins, destinations, routes, departure times and dwell at each stop. Household travel surveys ask a few thousand respondents to recall a single day every several years, and permanent count stations observe volume at a fixed point with no information about where a trip began or ended. A mobility panel covers both dimensions at once, continuously and nationally, at the cost of observing devices rather than people and inferring mode rather than recording it. That trade shapes most of what the field builds from it: expansion factors, origin destination matrices, disruption baselines, and access measures counting trips actually taken rather than trips theoretically possible.
How transportation researchers use mobility data
Expansion factors for traffic and pedestrian counts
Short duration counts, whether from tube counters, pedestrian sensors or a week of manual observation, have to be expanded to annual averages using temporal adjustment factors, and mobility panels can supply those factors for locations where no permanent counter exists. Researchers derive weekly, seasonal and hourly profiles from device traces near the count site and apply them to the observed sample, which matters most for walking and biking, where permanent count infrastructure is sparse. The panel is a sample of smartphone users rather than a count, so penetration varies by neighbourhood, mode and device type, and factors built this way still need validation against continuous counters before they carry a reported figure.
Latent demand and directional service gaps in transit
Planning that optimises against realised ridership cannot see the travel a network suppressed, so researchers use origin destination flows inferred from device traces as a demand side measure independent of the timetable, then compare it against scheduled service to find corridors and directions where supply lags. The asymmetries this surfaces, reverse commutes and off peak flows in particular, are invisible in farebox and automatic passenger count records, which register only the trips the system already carried. Mode is not observed in a raw trace: transit trips are inferred from speed, route geometry and stop proximity, and that inference weakens on corridors where a bus shares the roadway with cars.
Network disruption and recovery after extreme weather
Because a mobility panel runs continuously, researchers can set a pre event baseline of trip volumes, trip lengths and travel times for a county or a road segment, then measure how far activity falls after a flood, storm or closure and how long it takes to return, which produces resilience metrics comparable across places that were never instrumented in advance. The same design supports evacuation compliance and shelter in place estimates at a grain no post event survey reaches. The observation problem is that the shock degrades the data as well as the network: power outages and connectivity failures thin the panel exactly where disruption is worst, so a fall in observed trips confounds a fall in travel with a fall in reporting.
Mobility datasets available on Dewey
Veraset
Device level location pings with timestamps and coordinates, which transportation researchers reconstruct into trips, routes, departure times and stop durations rather than reading as visit counts.
View datasetAdvan
Foot traffic and mobility records that tie visits to individual places back to visitor home areas, giving origin side detail for accessibility, catchment and disruption baselines.
View datasetSafeGraph
A point of interest base layer with visit and spend measures, used to define the destinations that trips are assigned to and to attach activity type to an otherwise anonymous stop.
View datasetMore published work using mobility data
Callie Clark, Anne Driscoll, Xiyuan Ren, Salsabil Salah, Marta C. Gonzalez, Joseph Y. J. Chow, Takahiro Yabe · working paper · 2026
Youngjae Won · Cities · 2026
Kim, Sanghwa, Kannan, PK · Working Paper · 2025
.svg)