Modelling COVID-19 travel rebound with automated land use identification

Abstract

As movement restrictions during the COVID-19 pandemic forced urban workforces around the world to temporarily adopt telecommuting or flexible working arrangements, some speculate that these practices could remain as the ‘future-of-work’. Therefore, transportation and urban planners would both need to react to new post-pandemic work-based travel patterns. Unlike most common methods of analysing post-COVID telecommuting trends that rely on survey responses, this study develops a two-stage methodology of automatic land use identification (ALI) and mixed effects regression for the synthesis of both land use and transportation data with the aim of monitoring the post-pandemic travel recovery situation. Firstly, clustering methods are used for ALI around public transport destinations to generate different classes of regions based on land use characteristic. Mixed effects regression is then conducted to estimate the variability between different classes of regions. To gain insights on the travel rebound in Singapore, the case study focuses on business entity locations and bus transit volumes during the peak hours. Predictive modelling of a hypothetical travel recovery situation indicates that pre-COVID levels of traffic demand could likely return. The findings from this study have implications on transportation and urban planning, as well as decision-making in the post-COVID world and can be used as a basis for further COVID-related behavioural studies.