Analytics for Better Urban Cycling at the City of Toronto

Cycling is an effective way to promote healthy lifestyles, improve urban mobility, and combat climate change. However, safety and comfort concerns remain major barriers to cycling uptake globally. To make cycling safer and more inviting, we collaborated with the City of Toronto to develop and implement a suite of analytics tools for urban bike infrastructure planning. First, we propose a descriptive analytics framework that assesses cycling stress—the discomfort cyclists experience—on each road segment and quantifies the network’s low-stress cycling accessibility. This framework has been used to evaluate more than 10 bike infrastructure projects in Toronto and to estimate the infrastructure budget required to meet the city’s net-zero target. Building on this framework, we develop an optimization model that identifies optimal locations for new bike lanes. This model is a large-scale bilevel program that cannot be solved with existing techniques. In response, we develop a machine learning–augmented optimization method that is computationally tractable and can generate provably high-quality solutions. From 2022 to 2024, this approach would have matched the performance of Toronto’s implemented plan, reducing the required lane length by 25%, saving an estimated 18 million Canadian dollars. As of March 2025, 29.1 km of bike lanes recommended by our model have been approved by the Toronto City Council for construction between 2025 and 2027 with an additional 27.6 km under study for future implementation. Finally, we introduce a predictive model and a web application to support broader application of our methods. The predictive model assesses cycling stress based on street-view images, enabling deployment in cities lacking detailed road network data. The web application enables interactive refinement of optimization results, allowing practitioners to engage seamlessly with our tools. As urban centers increasingly prioritize sustainable transportation, this work offers a scalable and user-friendly framework for leveraging analytics to guide impactful infrastructure decisions.

History: This paper has been accepted for the INFORMS Journal on Applied Analytics Special Issue—2025 Daniel H. Wagner Prize for Excellence in the Practice of Advanced Analytics and Operations Research.

Funding: This work was supported by the Vector Institute, the City of Toronto, the University of Toronto Data Sciences Institute, and the Natural Sciences and Engineering Research Council of Canada Alliance [Grant 561212-20].