Date of Award

Spring 1-1-2017

Document Type


Degree Name

Master of Science (MS)


Applied Mathematics

First Advisor

Carson J.Q. Farmer

Second Advisor

Seth Spielman

Third Advisor

William Kleiber


The act of commuting to work is essential to many individuals, as it allows them to earn a living, to engage in their occupation, and to meet their essential needs. Continued research in transportation and urban commuting is necessary to the understanding of underlying influences on commuting patterns. A stronger understanding of the influences of these spatial interactions will lead to a more nuanced understanding of employment, housing, and the barriers associated with traversing the geographic distance between home and work. As such, it is important to examine the decision process behind a commute. However, commuting data, as with many spatial interaction data sets contains many structural and observational zero entries. To illustrate the utility of treating excess zeros in spatial interaction modeling as a byproduct of aggregate destination choices, this work develops, and assesses several different spatial interaction models of commuting flows within the contiguous United States. These models will demonstrate that the set of likely destinations available from a given origin is constrained, in large part, by distance as one might expect. Nonetheless, the implications of this assertion are much more nuanced than generally acknowledged in traditional spatial interaction models. The set of likely destinations for a given origin is largely influenced by distance, but in some cases, the actual commuting flows between origins and destinations reflect a decision-making process manifesting in a mixture of zeros and positive counts. As I will show, the models presented here provide a powerful and theoretically pleasing interpretation of commuting models, as well as a theoretical interpretation of the large number of zeros in many commuting datasets.