A49 Visualizing active travel sentiment in an urban context

Abstract

Abstract Background Commuting stress is known to be a root of many problems, including cardiovascular problems such as increased heart rate, increased blood pressure, back problems, as well as certain types of cancer. Examining how travel stress (or enjoyment) varies through time and space may reveal previously unknown relationships among active commute modes, intangible/tangible urban context, and daily weather. The present study set out to investigate these linkages using Geographic Information Systems (GIS) and space-time modeling techniques. Methods This research utilized exploratory spatiotemporal analysis and modeling of alternative commuter sentiment. Social media data was retrieved from twitter users in 6 metropolitan cities in the U.S., emotional content in the form of pleasure/arousal/dominance was extracted from it, spatial and aspatial data was obtained from various federal and local sources. A spatiotemporal weight was developed based on the space and time differentials for each daily tweeter. Exploratory analysis consisted of analyzing spatiotemporal patterning of tweeters using GIS techniques. Specifically, activity space analysis, surface estimation, spatial autocorrelation analysis (Getis-Ord hotspot analysis), global regression, and a spatiotemporal regression model was instituted. Results Preliminary exploratory analysis found that the majority of commuters tweeted once per day. When tweets were examined yearly, it was found that seasonality greatly affected commute sentiment patterns. Space-time weighted activity spaces varied significantly from un-weighted spaces. Furthermore, a weighted inverse-distance-weighted model demonstrated that travel mode affects human sentiment in both time and space. When hotspot analysis was visually examined in 3D, unique sentiments based on travel mode were observed in space and time. A global spatiotemporal regression model (OLS) indicated that weather, car usage, and mass-transit affected human sentiment (p-value Conclusions This analytical framework should be of interest to transportation planners seeking to understand how to promote alternative travel modes as well as health policy researchers studying the effect of commuting infrastructure on commuter’s well-being using a live, streaming dataset.