Estimating the Covid-19 Mortality Trend for Counties in the United States

Covid-19 impacted counties in the United States differently in the first surge phase (Mar-Jun 2020). There are counties that have very high mortality (New York, NY) while there are counties that have low mortality rates (Grand, CO). We want to know if there is any predictor that could indicate why a county has higher or lower mortality rates. We also want to see if the social vulnerability index and other demographic factors play a role in the change of Covid-19 mortality. We have aggregated multiple datasets to find the Covid-19 predictors for counties in the United States. We also clustered these features using hierarchical clustering with dynamic time warping distance metrics and k-means clustering with exponential regression. According to K-means and Hierarchical Clustering using Dynamic Time Warping, Covid Mortality is divided into 5 different clusters. Using Elastic Net, we can conclude that crowding, income, stringency, access to the nursing homes, political leaning, percentage incarcerated, and being over 65 and impact to a higher covid Mortality. Having lower education being obese, being Hispanic, and being more distant from airports impact to a lower covid mortality. Social Determinants of Health: transportation, political leaning, income, and other ones that we tested for ARE significant in determining excess mortality.