Using temporal public transport demand profiles to reveal urban spatial patterns

Abstract

We present a versatile method, inspired by computational neuroscience, for reconstructing smooth demand profiles from sparse timestamp data for public transport boardings. We show how areas can be clustered based on the similarity of their temporal demand profiles to reveal urban spatial patterns. We use the Helsinki metropolitan region to showcase the method using data on boarding events from the TravelSense data from HSL (the Helsinki region transport authority) collected through their mobile ticketing app. Our results show the show the dependence of travel demand on available public transit and modes and supply volume. Furthermore, the clusters align with extremely well with the types of urban areas in the region. Due to the high supply and even frequency of transit options, the differences in demand profiles are due to mode availability and land-use features rather than frequency patterns.