Quasi-Sparsity in Transportation Origin-Destination Demand

Term Start:

June 1, 2024

Term End:

May 31, 2025

Budget:

$120,000

Keywords:

Origin-Destination (OD) Demands, Quasi-Sparsity (QS), Transit, Travel Demand

Thrust Area(s):

Data Collection Mechanisms, Data Modeling and Analytic Tools

University Lead:

University of Washington

Researcher(s):

Xuegang (Jeff) Ban

Quasi-sparsity (QS) indicates that for a large-scale transportation network, most origin-destination (OD) demands are concentrated on a small fraction of the OD pairs, while majority of the OD pairs exhibit small (maybe non-zero) travel demands. One example is the King County network (the area that includes the City of Seattle in the State of Washington): more than 90% of the nearly 500,000 OD pairs in the network have less than 2 trips per day, and such trips count for less than 15% of the total demands in the network. While QS was proposed in Wang et al. [1] and used therein for the estimation of vehicular OD demands, its existence on real-world transportation networks and other modes has not been well studied, and its potential to help improve existing OD demand estimation methods (e.g., for vehicular demand matrices) or OD synthesis methods (e.g., for freight demand matrices) has not been explored. Many interesting questions still remain open: e.g., does QS hold for all the major modes (car, transit, freight) of a network? If so, how similar are the QS properties between different modes? Also, how does the QS property for the same mode on the same network evolve over time? Answers to these questions are not only scientifically intriguing but also helpful to the understanding of inherent human mobility patterns and to practical OD estimation/synthesis. This research aims to investigate the QS of OD trave demands on large-scale transportation networks, aiming to answer the above key questions. We plan to collect agency and open-source data, and will also leverage the aggregated big mobility data from third-party from previous research projects. Using the data, we will define formal measures of QS, and apply them to study the QS property of different networks and for different modes, including vehicular traffic, transit, ride-hailing services, and freight traffic. We will also compare the similarities of the QS measures for the OD matrices of different travel modes on the same network, and investigate the connections between the similarities and the inherent human mobility patterns of the network.

Scroll to Top