2021 Virtual Undergraduate Research Symposium
2021 Virtual Undergraduate Research Symposium
Capturing the Qualitative Behavior of the Spread of COVID-19 in Colorado Counties Using Merge Trees
Capturing the Qualitative Behavior of the Spread of COVID-19 in Colorado Counties Using Merge Trees
PROJECT NUMBER: 41 | AUTHOR: Amandin Chyba Rabeendran​, Applied Mathematics and Statistics
MENTOR: Tulay Flammand, Economics and Business
ABSTRACT
The primary objective of this work is to study the behavior of Colorado counties in correspondence to the spread of COVID-19. We break down the new daily cases per county and develop a methodology to compare such curves based on topological structures called merged trees. The motivation comes from the observation that spatial heterogeneity might lead to significant variations in the spread of COVID-19. Colorado is made of 64 counties, which largely differ in their population density per square miles. The diffusion of SARS-CoV-2 is studied to qualitatively assess similarities between pairs of counties, and classify them. We developed an algorithm to systematically compute the merged tree corresponding to the new daily cases curve for any given county of Colorado, as well as identified variables related to the merged tree to cluster the counties in a three dimensional space. The next will be to refine the clustering method and correlate the clustering to parameters associated to the counties.
PRESENTATION
AUTHOR BIOGRAPHY
Amandin Chyba Rabeendran is a sophomore at Colorado School of Mines (CSM) in the Applied Mathematics and Statistics department. This school year he conducted research on the spread of COVID-19 in Colorado and applied methods for comparing the behavior of different counties. He hopes to continue doing research within this field of applied statistics and topology even if it is not necessarily about COVID-19.
This is an interesting way to compare the evolution of the pandemic in different counties. If the trees cluster into two clusters I would be interested in seeing the spatial distribution of the different clusters. Is one cluster more common in metropolitan areas etc.. Nice job!
This is an overall very interesting project. Just an extension off of this, have you tried using any other classification algorithms besides clustering to see if any added information could be extracted from various algorithms?
Very nice research. According to this analysis, can you predict what would be the county behavior in the future?
Can you estimate what factors control different behaviors in different counties?
Really interesting work!
Nice work, Amandin! Could this kind of clustering be used to identify features associated with counties with anomalous spread (e.g., a low population density county than happens to contain a ski resort where transmission is high)?