Visualizations published in research papers and on the web
...
Currently (10 July), for an event with ten people in Albany County, there is a 4% chance that one of the people at the event has an active SARS-CoV-2 infection (see image below, from the COVID-19 Event Risk Assessment Planning Tool at Georgia Tech)
...
...
Visualizations from our working group
| Iframe | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
Visualizations from our working group
This post is dedicated to mapping reported incidence based on the confirmed number of cases (source). Below I have made some county-levels maps for the Continental US and the region around Wyoming. Here, I focus on “rolling” incidence over the past 3 days to highlights emerging “hotspots”. The 2010 census is taken as the population at risk of each county. I will provide updated figures every few days.
...
...
Mapping small counts can be problematic due to noisy values, delays in reporting, lack of testing, etc. Here, I use relatively simple spatial models based on county adjacency (sharing a border) to “smooth” (predict) incidence rates at the county level in the Continental US and in the region around Wyoming. These models now explicitly adjust for county rurality (measured by the 2013 USDA rural-urban continuum codes), elevation, median age, and the number of persons per household, so that the latent spatial effects (state- and county-level) are not reflecting any of those variables. Please see the attached R code for statistical model details.
I have pivoted to modeling the incidence rate over the last 7 days to make a prediction of “current” risk:
...
...
R code for mapping and analysis. Code has been updated to download data directly from GitHub, which is updated daily. Files for the US Counties base map and the 2013 USDA county rurality codes are provided for reproducibility:
...