Analysing transit hub patronage: Meta-clustering using smartcard data

Understanding the long-term demand and functional dynamics of public transport hubs is crucial for effective urban planning and transport service operations. This paper presents a comprehensive data-driven method for analysing public transport demand, leveraging over 480 million smartcard transactions across 44 transport hubs in Perth, Western Australia, over a period of four years (2016–2019). We propose a novel multi-stage clustering approach that synthesises diverse data perspectives-ridership trends, activity profiles, passenger demographics, and inter-hub connectivity-into a cohesive understanding of transport hub performance. The hubs are characterised by the patterns of activities inferred from ‘stays’. Our methodological contribution includes the development of a meta-clustering technique that integrates various clustering results into a consensus framework, enhancing the robustness and interpretability of the findings. The final meta-clustering solution revealed aspects that could not be seen in the individual clusters. Practical contributions of this study offer granular insights into the temporal and spatial patterns of hub usage, facilitating targeted interventions for transport and urban planners. While several hubs had an increase in patronage, others experienced reductions or remained relatively stable. Patterns can be tracked to specific types of activities, land-use, or infrastructure changes. Results indicate that hubs with Generator and Connector functions, served by substantial Park-and-Ride and feeder bus services, demonstrate resilience to external changes, whereas Attractor hubs are more susceptible to declining patronage in response to macro-level shifts. This study provides a methodological blueprint and actionable knowledge for transport authorities to optimise public transport systems, ensuring they meet the evolving demands of urban populations.