This thesis presents machine and statistical learning approaches for sustainable planning in infrastructure and mobility systems. First, I have developed a convolutional neural network (CNN) to predict tree failure likelihood. Such assessments have traditionally been performed manually. I conduct a visual analysis of the predictions, indicating an approach for incorporating interpretability into model selection. Benchmarking the results against those produced by state-of-the-art CNNs, I show that a relatively simple model produces better results in a computational time that is three times faster. Via this novel framework, I demonstrate the potential of machine learning to automate and consequently reduce the costs of tree failure likelihood assessments in proximity to power lines, thereby promoting sustainable infrastructure. Secondly, I examine the effects of COVID-19 on mobility, segmented by transportation type, as well as social activity such as workplaces and residential, and their interdependencies. Using time series data across five continents, I estimate a Bayesian global vector autoregression model which explains patterns in activity and mobility trends and analyze their relationship with COVID-19 spread. I expect that the model framework and outcomes will guide policymakers to adopt appropriate measures to mitigate and safely recover from future disease outbreaks.
| Identifer | oai:union.ndltd.org:UMASS/oai:scholarworks.umass.edu:masters_theses_2-2482 |
| Date | 01 February 2024 |
| Creators | Apostolov, Atanas |
| Publisher | ScholarWorks@UMass Amherst |
| Source Sets | University of Massachusetts, Amherst |
| Detected Language | English |
| Type | text |
| Format | application/pdf |
| Source | Masters Theses |
| Rights | http://creativecommons.org/licenses/by-sa/4.0/ |
Page generated in 0.0684 seconds