Mining urban passengers’ travel patterns from incomplete data with use cases

Abstract

The rapid development of the Internet of Things (IoT) and Intelligent and Connected Transportation Systems (ICTS) are making our city smarter and greener. For large cities with millions of population, their public transit systems are of great significance to mitigating the road congestion along with reducing the emission of greenhouse gases. One critical problem transit authorities encounter is that they can not clearly understand the actual behavioral preference and travel demand of their passengers, worse even, nowadays, the passively collected data from IoT devices do not guarantee the integrity of information and make it more difficult. To address these problems, in this research, we first propose a novel framework to derive passengers’ closed transit chains along with their home and work locations from incomplete travel records using an information enrichment and probabilistic inference approach. We then leverage both evaluation and volunteers’ records to evaluate the usability and theoretical boundaries of our methods. We finally apply our proposed framework to mine a series of useful information about the city and behavioral preferences of our passengers. Our proposed methods are applicable to mining individuals’ behavioral patterns from sparsely collected crowdsourcing data in future.