Analysis and clustering of massive public transport data

Abstract

This work proposes a methodology for analyzing citizens’ movement patterns in big cities by exploiting validation records from tra- veler cards in public transport networks. One important question that naturally arises about traveler movements is how many people travel across a line in a given time frame, that is, line load. However, answering such a simple question by exploiting the traveler card records in a traditional database proves costly both space and time-wise. In this work, we present a novel representation of boarding data that leverages Functional Data Analysis to efficiently answer queries about average traveler load for a given stop, day type, and time frame within the day.