Friday, May 29, 2020
10:00 AM
Zoom Meeting ID: 674-436-8651
Ph.D. Thesis Defense
Brian Dickinson
University of Rochester
Improving our Understanding of Society through Global Patterns of Human Mobility
Many dynamic systems which are simultaneously difficult and critical to understand are driven by the actions and reactions of thousands or even millions of individuals. Such areas of interest range from the urban planning of individual cities to issues of global economics, migration, and public health. In many of these areas, existing statistical measures focus not on the individuals driving the system, but rather on the physical infrastructure of the system itself roadways, public transit, bridges, buildings, and hospitals. While these metrics have proven exceptionally useful, their selection was likely based on the availability, as opposed to the suitability, of these indicators.

We show that revolutions in the production and availability of individual location data at a global scale can improve our understanding in all of these areas. In particular we show that, at the city level, mobility metrics are more correlated with transportation utilization, pollution, and public health than traditional measures such as population density and urban sprawl. At the neighborhood level we demonstrate that mobility provides insights into how the particular dynamics of different cities alter the movement patterns of different socioeconomic and demographic groups. The ways in which these patterns differ between cities and over time may well provide valuable insights in urban planning. Meanwhile at the global scale we show that mobility patterns can be used to infer nighttime satellite imagery a gold standard for uniform global analysis. Such satellite data has been used in applications from poverty mapping to climate change. We show not only that high fidelity artificial imagery can be inferred from mobility patterns, but that this imagery is often a more effective estimator of GDP than the original data. In all of the above applications we demonstrate the utmost regard for the preservation of individual privacy.

Advisor: Prof. Henry Kautz (Computer Science)

Committee: Prof. Jiebo Luo (Computer Science), Prof. Ehsan Hoque (Computer Science), and Prof. Gourab Ghoshal (Physics)