Jacobs, Ian Michael
(2024)
Exploring the Additive Effects of Religious Participation on Multivariate, Demographics Based Machine Learning Models.
Master's Thesis, University of Pittsburgh.
(Unpublished)
Abstract
Through the 21st century, vaccine hesitancy has had a significant effect on the implementation of vaccine development and rollout in the United States. A known and well documented factor that contributes to this kind of structural hesitancy is regular participation in a religious congregation or community whose doctrine or teachings condemn vaccination and/or modern medicine in some form. The public health contribution of this thesis is to support the use of machine learning in the prediction of public health outcomes, as well as promote the contribution of socially anchored metrics within demographics-based models.
Data for this project was sourced from The Department of Health and Human Services Office of the Assistant Secretary for Planning and Evaluation, The U.S. Department of Agriculture’s Economic Research Survey, and The Association of Statisticians of American Religious Bodies’ U.S. Religion Census. These data were cleaned at the U.S. county level and the remaining variables were categorized into six major demographic categories: education, population, poverty, unemployment, vaccine hesitancy, and religious participation. This cleaning process resulted in 54 usable demographic variables and one outcome variable.
After data cleaning was performed, four machine learning techniques were implemented on the variable set to compare their prediction ability: elastic net, multivariate adaptive regression splines, random forest, and gradient boosted trees. Using the root mean square error and R-squared of each of these models, it was determined that the gradient boosted trees method had the greatest prediction ability with this particular dataset.
Variable selection was performed, and it was determined through importance testing that 26 of the 54 variables had a significant contribution to the model and provided the most substantial prediction ability. Of those 26 variables, two originated from the religion category. Results from the gradient boosted tree analysis indicated a decrease in prediction ability when the selected religion variables were removed from the model, which supports a data-based linkage between vaccine hesitancy and religious participation. Post-hoc hierarchical clustering was performed at a county level to give a visual representation of the demographically constructed clusters and to provide a geographically based comparison between the selected demographics and vaccine hesitancy.
Share
Citation/Export: |
|
Social Networking: |
|
Details
Item Type: |
University of Pittsburgh ETD
|
Status: |
Unpublished |
Creators/Authors: |
|
ETD Committee: |
|
Date: |
14 May 2024 |
Date Type: |
Publication |
Defense Date: |
3 April 2024 |
Approval Date: |
14 May 2024 |
Submission Date: |
23 April 2024 |
Access Restriction: |
No restriction; Release the ETD for access worldwide immediately. |
Number of Pages: |
75 |
Institution: |
University of Pittsburgh |
Schools and Programs: |
School of Public Health > Biostatistics |
Degree: |
MS - Master of Science |
Thesis Type: |
Master's Thesis |
Refereed: |
Yes |
Uncontrolled Keywords: |
N/A |
Date Deposited: |
14 May 2024 19:04 |
Last Modified: |
14 May 2024 19:04 |
URI: |
http://d-scholarship.pitt.edu/id/eprint/46259 |
Metrics
Monthly Views for the past 3 years
Plum Analytics
Actions (login required)
|
View Item |