Modelling trip scheduling decisions of bus commuters amid disruptive events using smart card data

Departure time models are key tools for understanding time-varying travel demand. Nonetheless, there is limited research focusing on the analysis of trip scheduling decisions in the context of public transport users. In particular, research on how public transport users adapt departure times when the activity and travel landscape are altered as a consequence of disruptive events (e.g. pandemics, social unrest), is yet to be conducted. Smart card data, which passively records time-stamped departure locations of public transport users, offers the opportunity to investigate such shifts in detail but is yet to be utilised. The paper aims to address these two gaps by using smart card data to investigate the trip scheduling decisions of bus commuters amid disruptive events. This goal is achieved by estimating departure time choice models (DTCMs) for characteristic episodes between 2019 and 2022 for Santiago's bus system, a city affected to different degrees by two types of disruptive events within this timeframe: the COVID-19 pandemic and social unrest. The paper addresses the methodological challenges of calculating schedule delay with smart card data by estimating preferred arrival times as a random variable within a mixed multinomial logit model. The approach is assessed through the valuation of the trade-off between travel time and schedule delay (TVSD), with the results falling within the range of values previously reported in the literature. The model results highlight the existence of multi-temporal differences in the arrival time preferences of bus commuters, as well as in their TVSD amid disruptive events. It was found that bus commuters were less willing to accept an increase in their travel time to reduce their schedule delay during disruptive episodes. The heterogeneity between bus travellers was also explored: recurrent bus commuters exhibited higher TVSDs than occasional commuters. The outcome of this study supports using smart card data as a feasible source to investigate how public transport passengers allocate their trip scheduling both during normal periods and amid external disruptions.