Characteristics of cyclists using fitness tracker apps and its implications for planning of bicycle transport systems

The surge of crowdsourced cycling data from various fitness tracker apps has attracted the attention of transportation planners as it has the potential to provide rapid, cheaper, and high-resolution data for the planning of cycle infrastructure. However, much has been asked on whether the data coming from these fitness tracker apps represent the demographics and trip characteristics of the total cycling population. To explore the question, this study conducted a field intercept survey of 320 cyclists in the city of Ann Arbor and Grand Rapids located in Michigan, United States. The cyclists were categorized into three main groups based on the reported fitness tracker usage namely cyclists using the Strava app, cyclists using other fitness tracker apps and cyclists reported not to use any fitness tracker app. The logistic regression and Bagging (Bootstrapped Aggregating) decision tree were used in the analysis. The survey results suggest that cycling data on cycling activities from the Strava app and other fitness apps can jointly represent different cycling populations by age and gender. As for the trip purpose, a cyclist who reported using cycling fitness tracker apps had higher odds of being on a recreational trip as opposed to a commute trip. This indicates that the fusion of different crowdsourced fitness tracking data alone cannot offset the bias in the trip characteristics despite the observed diversity in demographic characteristics. Consequently, the study emphasizes the need for complementing crowdsourced cycling data emanated from multiple fitness tracker apps with other data sources to capture the travel behaviors of both commuting and recreational riders.

Automated summary

The study found that cycling data from fitness tracker apps can represent different age and gender groups within the cycling population. However, even with the observed diversity in demographic characteristics, simply merging different crowdsourced cycling data still presents biases. Therefore, supplementing with other data sources is necessary to capture the travel behaviors of both commuting and recreational riders.