A Clustering-Based Framework for Individual Travel Behaviour Change Detection

Abstract

The emergence of passively and continuously recorded movement data offers new opportunities to study the long-term change of individual travel behaviour from data-driven perspectives. This study proposes a clustering-based framework to identify travel behaviour patterns and detect potential change periods on the individual level. First, we extract important trips that depict individual characteristic movement. Then, considering trip mode, trip distance, and trip duration as travel behaviour dimensions, we measure the similarities of trips and group them into clusters using hierarchical clustering. The trip clusters represent dimensions of travel behaviours, and the change of their relative proportions over time reflect the development of travel preferences. We use two different methods to detect changes in travel behaviour patterns: the Herfindahl-Hirschman index-based method and the sliding window-based method. The framework is tested using data from a large-scale longitudinal GPS tracking data study in which participants had access to a Mobility-as-a-Service (MaaS) offer. The methods successfully identify significant travel behaviour changes for users. Moreover, we analyse the impact of the MaaS offer on individual travel behaviours with the obtained change information. The proposed framework for behaviour change detection provides valuable insights for travel demand management and evaluating people’s reactions to sustainable mobility options.