Bike-sharing station profiles and urban context: Understanding imbalance with asymmetry-based clustering

Abstract

Depicting the usage characteristics of a bike-sharing system can contribute to more efficient planning and operation. Station-level imbalances are typically treated as operational defects to be corrected. However, the systematic asymmetries between pick-ups and drop-offs may reflect stable behavioural and spatial patterns. The study aims to reframe imbalance as a behavioural and spatial signal that can be systematically profiled. This paper proposes an integrated framework to analyze station-level asymmetry in a station-based bike-sharing system and to derive interpretable behavioural station profiles. The framework combines four steps: temporal analysis, calculation of absolute and relative asymmetry indices, asymmetry-based K-means station clustering into specific profiles, and GIS-based mapping of profiles. The approach is applied to trip records from MOL Bubi public bike-sharing system in Budapest. The results reveal clear contrasts between weekdays and weekends, as well as between weekday commuting peaks, and robust source–sink patterns between the inner city and surrounding residential areas during weekdays. Five behavioural station profiles are identified, capturing urban contexts. Stations in residential areas act as sources in the morning and sinks in the afternoon; stations in downtown areas act oppositely. Spatial analysis highlights recurring features, including changes in usage patterns near railway and metro stations. By explicitly linking asymmetry-based station profiles to urban structure and public transport interfaces, the framework supports data-driven planning, rebalancing strategies, and station placement. The Budapest case study provides new empirical evidence from an underrepresented Central and Eastern European context, demonstrating that asymmetry-based profiling yields stable and operationally relevant insights.