license.txt

Abstract

The recent shift toward urbanization has resulted in considerable speculation about the future of smart cities. Although the world population is projected to grow, researchers believe that the total rural population has already peaked as more people move toward cities. This expectation for increased urban population has called for the development of new technology to help make cities more e cient and capable of handling their increased number of residents. This thesis attempts to better explain this recent urbanization by modeling how people value proximity through analyses of trip distribution data and residential real estate pricing in the United States. Simulated nationwide trip data is examined under the lens of the gravity model to derive the the travel cost F matrix. This relative cost of travel is then regressed against current local residential real estate prices in attempt to model how much people are willing to pay to reduce their distances traveled during their daily trips. An extension of this model that this thesis also explores is determining optimal trip production reallocation for a geographic area such as New York City to minimize overall cost of proximity while still satisfying currently observed transportation demands by creating a convex optimization problem. Finally, this thesis attempts to model nationwide real estate prices using the transportation data as an input through elastic net regression and artificial neural networks. The findings by this thesis could help provide useful information to the developers of the next smart cities, both through improved urban planning and zoning and the development of new smart-city technologies such as repositioning fleets of autonomous taxis.