Understanding the role of street network configuration on commuter cycling can aid city planners in assessing and evaluating interventions to promote regular cycling into people's routines. Studies examining this relationship generally build models on observed cycling counts. While this provides valuable information, it could still result in an incomplete picture because some routes with advantages (e.g., good accessibility, amenities etc) may not have been utilised as expected because of other confounding factors (e.g., safety issues, motorised traffic) or vice-versa, leading to incorrect conclusions especially in the absence of data on these confounding factors. Thus, in this study, we argue that observing higher cycling flows on a route should not be the sole criteria to examine the role of street network configuration on the cycling patterns, especially when certain confounding factors have not been controlled for. We utilise data from the activity tracking app Strava for the city of Glasgow and compare the observed cycling intensities with the modelled cycling intensities where all the cyclists take the shortest routes. We estimated three linear regression models for: Observed Strava Cycling Intensities, Modelled Strava Cycling Intensities, and the difference between these two measures. Street network configuration were incorporated using Space Syntax measures: Normalised Angular Choice and Normalised Angular Integration. The roles of these variables, along with route characteristics and natural environment factors, on commuter cyclists' trips are explored. Visual exploration and linear regression models indicate that cyclists deviated away from the well-integrated, straighter routes to aesthetically attractive routes with cycling infrastructure, and towards links within mixed land use and to the least deprived areas. These results are of interest to policy makers and assist in infrastructure planning.
Read full abstract