Models for estimating zone-level bike kilometers traveled using bike network, land use, and road facility variables

Despite the increase in studies that link bike trips with various correlates, research gaps remain, including a lack of empirical tools to predict bike kilometers traveled (BKT) using comprehensive measures. The present study evaluates the impacts of network indicators, land use, and road facility on BKT by developing zone-level ridership models. Land use and road facility data were collected for 134 traffic analysis zones (TAZs) in the City of Vancouver, Canada. In addition, graph theory was used to obtain bike network indicators for each TAZ, including measures of connectivity, continuity, linearity, slope, and length of the bike network. A full Bayesian approach, accounting for spatial random effects among the TAZs, was used to develop the models. The results suggested that more connected, dense, flat, continuous, recreational, and off-street bike networks yielded higher BKT. Models that accounted for spatial effects were found to have better fit than those that did not incorporate spatial effects, which implies the importance of considering spatial effects while modeling BKT at the aggregate level. The models provide insights about the factors that influence BKT and information about the spatial variability of the bike travel within a city, which can be useful for planning bike networks.