Long-Term Trends in Activity Participation Observed with Crowdsourced Data from an Experiment with the “OneBusAway” Application

Abstract

We discuss the possibility of obtaining long-term travel behavior information through an open-source, crowdsourced, low-energy data collection effort. We utilized the “OneBusAway” travel planner that is used in several regions of the United States to obtain data from occasional and frequent transit users. From those willing to participate, locational data were obtained via Google’s Android Activity Transition Application Programming Interface that collects records as to when the user is changing their mode of movement among the possible states (still, walking, running, cycling, in vehicle). At a transition point the time and location were recorded. We discuss data completeness problems and an approach to clean the data to obtain continuous 24-h records of those users with missing elements during days marked as “unknown.” We further infer the home location of respondents based on the locations the volunteers primarily were located at night. Data were collected from 7,563 users, some of whom provided data for the whole period. We show trends in time spent in-vehicle, walking, and remaining still. Our findings suggest that, considering all activities, the COVID trends were less pronounced. Further, after COVID we found a rebound effect, followed by a decline in activity. In 2023 our sample, especially the residents of Seattle, spent more time being mobile again. The data further suggest that to quantify behavioral trends, it is important to distinguish those presumably experiencing a shift in behavior from those who are likely to have “settled into a pattern.”