Rebecca Youngerman
Data Scientist, Public Health Enthusiast, Looking to Change the World
Graduate School Work
COVID-19 Work
In Spring 2020 the world took a sharp turn to focus on an unprecedented challenge: COVID-19. As a graduate student in Health Data Science, my professors and I saw the importance of analyzing this pandemic in real-time. Three of my courses: Statistical Learning, Data Science II, and Computing Foundations led to final projects looking to answer various questions about Coronavirus:
1) How does socio-economic status interplay with the spread of COVID-19 in New York City?
2) Can we use lung CT scans to predict COVID-19 as an alternative to potentially inaccessible testing?
3) How can parallelization improve the efficiency of simulating the spread of a disease like COVID-19?
Health Financing Analysis
Using data from the Gapminder Foundation I created an interactive app with R shiny. The analysis explores the variance in health outcomes and financing around the world through interactive maps and country-by-country comparisons. Have fun exploring!
See the full, collaborative project here: https://rebeccayoungerman.wixsite.com/healthandwealth
Academic Posters
Undergraduate Work
Honors Thesis
This paper details my self-designed project through AidData and GeoQuery working within a mission to create specialized GeoQuery platforms addressing the United Nation's Sustainable Development Goals. This project embarked on analyzing and publicizing sub-national data focused on SDG 3.C (found here) in Kenya.
Policy Brief
For the culmination of Kinesiology 460, Health Policy, taught by Professor Carrie Dolan, PhD at the College of William and Mary, student were tasked with creating a 500-800 word policy story that incorporates key aspects of data journalism. The story needed to adopt an evidence-based approach to critically analyze the following question: "What health issue should be addressed to reach sustainable development goal number 3?" (see goal description here).
Child Deprivation Index
This report addresses an issue posed by Save the Children. The organization wants to create a Child Deprivation Index that can be standardized across countries to aid in decision-making about where to do direct programming, with a focus on the most deprived children and groups. Save the Children proposes to use a 3x3 index based on under 5 mortality rate, years of schooling completed (or similar education metric), rate or stunting among under 5s, and location (urban/rural), gender, and household wealth.
Dam Prioritization
This report examines the question: "Given expert practitioner preferences, and accounting for uncertainty in your source data, what 10 dams would you recommend for a priority investment by the US Army Corps of Engineers?"
In order to answer this question, a predictive model of risk based on past dam failures needed to be created, selecting the best model for prediction from a suite of potential models. Then, a mixed qualitative-quantitative model was created using the Analytic Hierarchy Process to select, of the top dams most likely to fail, which should receive priority investment.
College Loan Default Rate
This report details a model that will predict the probability of student loan default for US academic institutions. The top-5 best schools a student should go to if they want to minimize the likelihood of defaulting on student loans post-graduation based on the student's stated preferences was then provided based on the model. Web scraping, data joining using a fuzzy matching procedure, and Monte Carlo cross-validation were all used to create a final recommendation of the five best and five worst universities.
Predicting Syrian Conflict
This report focuses on integrating disparate data sources in order to fit a model predicting conflict in Syria. The analysis was done at two geographic scales using different census units in order to contrast results and obtain a more accurate model. Web scrapped data from Google Maps was combined with conflict data to create models. Parallelization was necessary within the simulation code was used to ensure a solution could be found in the time allotted.
Geospatial Analysis:
Projects in ArcGIS
.png)
.png)
.png)
