Improving the understanding of travel behaviour over time

Travel behaviour changes as people’s circumstances, preferences and transport systems evolve. Understanding these changes helps explain shifts in demand, evaluate policies and plan transport services. Ideally, behavioural dynamics would be analysed using panel data that follow the same individuals over time. In practice, however, evidence comes from data sources that differ in granularity, temporal structure and volume. Some sources provide grouped market shares or spatial summaries, while repeated cross-sectional surveys observe different individuals at each wave. Passive sources generate detailed records across space and time, but their size and complexity can make meaningful patterns difficult to identify, and they often provide limited information about the individuals they represent. Each source preserves different dimensions of travel behaviour or demand while obscuring others, supporting different levels of inference.

This thesis addresses these challenges through three papers. The first investigates the consequences of estimating choice models from aggregated data. Using a controlled simulation, it evaluates the bias and precision of grouped models and identifies the conditions under which they can recover parameters and behavioural indicators obtained from individual data. The second examines temporal heterogeneity in travel preferences using repeated cross-sectional surveys from Bogotá. It estimates a group-level latent class choice model with switching class allocation to identify changes in modal preferences and behavioural profiles across zonal groups over time. The third provides an empirical analysis of multi-year smart card data from Bogotá’s Bus Rapid Transit system. It applies Functional Data Analysis methods to characterise daily station demand profiles, assess their spatial patterns across the network, and examine temporal changes.

Together, the papers show how to analyse travel behaviour and demand across data sources with different structures and limitations. They contribute to transport research, travel behaviour analysis and choice modelling, while expanding empirical evidence on long-term travel behaviour and public transport demand in Bogotá, a large Latin American city where individual panel data remain scarce.