Modeling cyclists speed at signalized intersections: Case study from Ottawa, Canada

The study of cyclist behavior for developing realistic and reliable behavioral models is attracting research focus. Achieving a detailed understanding of cyclist behavior is a cornerstone in building micro-simulation models and ultimately creating a more sustainable transportation system. Cyclist behavior is especially important at traffic intersections due to the exposure to turning and crossing vehicle movements. This study focuses on cyclist speed modelling at traffic intersections in urban areas. Data collection was conducted at three different traffic intersections in the downtown area in Ottawa, Canada. Video monitoring covered cyclists, vehicles, and pedestrian movements. Cyclists approached these intersections through physically segregated bike lanes. Cyclist speed was measured based on metric measurements at the intersections and temporal measurements from the video data. The variables associated with cyclist speed that were examined are: pedestrian crossing movements, adjacent vehicle traffic, traffic signal indication, type of right-turn lane, potential conflicts with turning vehicles, and occurrence of traffic violations. Multivariate regression analysis was conducted to link cyclist speed and the explanatory variables. The resulting R
²
values were 0.43 for predicting cyclist crossing speed and was 0.56 for predicting the change in cyclist speed while approaching the intersection. A number of statistically significant associations were observed and documented. Overall, considering the data collection effort, sample size, and the predictive power of this regression model, it can be concluded that the developed models are of potential practical use in predicting cyclist crossing speed.