Business Analytics Seminar: Farnaz Farzadnia, Copenhagen Business School

Presenter: Farnaz Farzadnia, Copenhagen Business School

Title: Safety in Bicycle Lanes: An Explainable Approach

Abstract: Cycling is an important part of sustainable urban mobility but ensuring safe and comfortable bicycle lanes remains a key challenge for urban planning. This work develops data-driven methods to assess and explain bicycle-lane safety at both the segment and network levels. First, we address the challenge of updating safety-related clusters while maintaining consistency with prior labels and explaining any changes. A multi-objective optimization approach balances cluster quality, consistency with prior knowledge, and rule-based explanations at both the segment and cluster levels. Second, we estimate bicycle crash risk for different crash types and identify the road features contributing to these predictions through global and segment-level explanations. To support network-level planning, we group road segments into homogeneous cohorts based on their risk profiles, providing an interpretable basis for targeted safety interventions. Using data from Copenhagen and Frederiksberg, Denmark, our results demonstrate how these complementary explainability approaches can provide transparent and actionable support for bicycle-lane safety planning.

Host: Marcel