DATA DRIVEN DEVELOPMENT: ANALYZING URBAN CHANGE THROUGH CONSTRUCTION, DEMOLITION, AND RENOVATION

A communitys urban form is determined by the interaction of transportation networks and adjacent land uses. This relationship has been used to understand and measure a variety of development patterns including urban sprawl with its low population density, segregated land uses, and auto-centric street network. A deeper understanding of development patterns is critical to understanding how cities change and how that change may affect the people residing in those communities. Construction, demolition, and renovation are key indicators of this change and while varying measures of each have been used to gain insight into how communities are changing, they are typically measured indirectly or in isolation which fails to paint a comprehensive picture of what is occurring in a community. This dissertation addresses this critical gap by first treating construction, renovation, and demolition as direct agents of urban change and defining the location and timing of this change using standard administrative permit data. Furthermore, it expands upon current research by analyzing 122 variables using a series of random forest regression models to predict rates of construction and demolition and then identifying the characteristics that are most closely associated with each. A series of models that test prediction accuracy for construction and demolition independently were outperformed by a model that used a net construction score which combines construction and demolition together into a single measure. This model identified a combination of landscape and socioeconomic variables such as the number of patches, the percent of residents living at or below poverty, and the percent of non-white population as having the greatest impact on determining rates of net construction. These models were also tested against a series of metrics relating to local governments role which was found to be negligible. Cities are complex entities and to understand the factors that contribute to their evolution requires methods and techniques that are able to better represent this complexity. The work outlined in this dissertation presents such a methodology, one that is able to explore urban change holistically while retaining the complexity that exists between the multitudinous variables affecting community change