Machine Learning for Public Policy: Applications in Infrastructure and Air Pollution
thesisposted on 19.11.2020, 19:17 by Samuel Jones
While machine learning has become ubiquitous in certain fields, its public policy applications in the infrastructure and air pollution domains are relatively understudied. This work develops two frameworks to analyze (1) the impacts of infrastructure and social equity and (2) the impact of area
source pollution on resultant concentrations to better inform public policy along these two domains. This work employs a range of machine learning techniques to perform variable search and selection and analyzes causal connections between infrastructure (i.e., bridges) and social equity factors. Further, a novel neural network design that combines vector autoregression techniques with pollution data capably predicts pollution concentrations for two species of PM2.5. Throughout this work, the techniques and frameworks are specifically designed to be accessible by
engineers and policymakers in the infrastructure and air quality managements domains.
DepartmentEngineering and Public Policy
- Doctor of Philosophy (PhD)