Distance Decay and Commuting Peaks: Evidence from Mining Transit Smartcard Data in Canberra, Australia

Abstract

Big data manifests itself in the public transport arena through smartcard ticketing systems. Although primarily designed as a fare collection mechanism, some systems embed important features like alighting points which add an additional layer of analytical capability. Mining this data can reveal travel behaviour with census-level accuracy not previously possible through travel surveys. One such application is in the exploration of spatial-temporal variations in transit patronage, an omnipresent concern for transport planners seeking to meet peak demand. To date, there is yet an understanding for how commuting peaks are influenced by the friction of distance, a quintessential geographic concept with widespread applications from transport network design to the urban form of sustainable cities. Data was sourced from a high quality smartcard system at a medium-sized bus operator in Canberra, Australia. A 163 working day year-on-year comparison was undertaken to compare the before and after effects of a change in bus route structure and timetables. An open-source database and the R programming language and environment were used to produce a series of visualisations culminating in a loess regression. Temporal variations across the diurnal cycle were quantified through the peak-to-base ratio, which was plotted over the spatial variable, accumulated journey time. Unlike the hypothesised positive exponential relationship, the analysis revealed a pseudo-sinusoidal curve reaching maximum peaking intensity for journey lengths around 30 minutes. This held for adults and school students, but seniors and tertiary students exhibited vastly different curves better resembling the hypothesis. Minimal effects also arose from the network change, suggesting that the results are somewhat entrenched in human behaviour. This thesis offers important theoretical, methodological and practical contributions. A distance decay function has, for the first time, been drawn for commuting peaks, possible only though and reaffirming the importance of embedding alighting points in transit smartcard systems. The findings also endorse the merits of compact cities, as shorter journeys are linked with the least magnitudes of peaking, thereby permitting the best use of transportation resources. Show more