Mining temporal and spatial travel regularities for transit planning

Smart Card data from Automated Fare Collection system has been considered as a promising source of information for transit planning. However, literature has been limited to mining travel patterns from transit users and suggesting the potential of using this information. This paper proposes a method for mining spatial regular origins-destinations and temporal habitual travelling time from transit users. These travel regularity are discussed as being useful for transit planning. After reconstructing the travel itineraries, three levels of Density-Based Spatial Clustering of Application with Noise (DBSCAN) have been utilised to retrieve travel regularity of each of each frequent transit users. Analyses of passenger classifications and personal travel time variability estimation are performed as the examples of using travel regularity in transit planning. The methodology introduced in this paper is of interest for transit authorities in planning and managements