Yearly patterns of transit usage: a cumulative clustering approach using 7 years of smart card data

Abstract

This paper presents results of the application of clustering methods to report on the yearly patterns of transit ridership, at metro stations and on bus lines, as observed using 7 years of smart card validations (2015-2021). Results from the k-means approach are presented in this paper while various methods were tested using the same vectors of data. The results confirm the ability of the proposed approach to understand the global changes in ridership patterns. The cumulative clustering approach used allowed to understand the evolution of both bus and metro across the seven years from a temporal and spatial perspective.