Nov 19, 2020 Update: I think it is time to retire this website, at least for now. However, there’s another website that just started publishing estimates that are more detailed and probably better than mine: https://covid19-projections.com/. So go check them out!
More on why I’m retiring the site: I have not been posting updates lately, for a variety of reasons. But the most important one is that the testing situation has changed enough in the US since I first developed my estimation approach that I’m not confident my method will still work well with the current data being released. So I was delighted to find that the website above has developed some ways of dealing with these new concerns I have about the data. Meanwhile, I’ve had to redirect my attention back to my “day job” and thus haven’t been able to devote much time to COVID-19 data lately.
The number of confirmed COVID-19 cases is a an important data point for anyone tracking the severity of the coronavirus pandemic. Unfortunately, trends in confirmed case counts can be misleading because these counts only reflect the number of cases that have been detected. Thus, an increase in the number of reported cases could mean that the number of infections has increased, or it could mean that we’re now doing a better job of detecting whatever infections are out there (or both).
The main purpose of this website is to provide estimates of disease prevalence that account for both the number of confirmed cases and the number tests that are being conducted. By taking into account how much testing is being done, we should get a more complete picture of whether infections are really going up or down.
One important limitation: the estimates on this site probably track with the number of infections occurring roughly 2-4 weeks prior to the date shown, not the number of infections that are currently occurring. The reason for this is that there are typically several delays between the time when someone first gets infected and the time when their case might get confirmed and show up in my estimates. Still, the number of confirmed cases allows for a more up-to-date view into outbreak severity than data on hospitalizations or deaths, which tend to lag even further behind.
These estimates rely on data reported by individual states (as assembled by The COVID Tracking Project), and while some basic checks for data consistency have been made, there are almost certainly further errors and inconsistencies in the data that are not flagged.
[Graphs last updated October 19, 2020]