Data and Delta

Changing to weekly data

Covid Proxima was conceived as a quick experiment, a way to poke at the New York Times’ dataset. I didn’t want to build hosting for this and wanted it to just run in a browser, so I could use it as a local tool and post it as a static website with no dependencies. It has all the delightful messiness of any quick and dirty side project.

Along the way, COVID-19 kept happening. Thanks to delta, the daily data dump is bigger than ever. So, to keep this site working on all kinds of browsers and CPUs, I just made a change to only compute data in 7-day chunks, moving backwards from the current day. Not perfect, and some of my previous posts read a bit strangely now, since they discuss daily results. I’ll fix those over time, but this was the quickest way to keep the startup time and memory usage down.

So, how bad is delta?

The impact of delta is remarkable. Cases are more than halfway back to our all-time peaks over the winter, despite a majority of American adults having at least one vaccination.

Unfortunately, we know deaths are a trailing indicator of cases. The US fatality rate remains at a relatively constant 11 out of 1,000 (1.1%), so expect the deaths to go up quite a bit in the coming weeks. Per capita, the most infections are happening in Mississippi, Louisiana, Florida, Alabama, and Arkansas, with all of them near or above all-time weekly infection highs.

Since Florida isn’t reporting county-level death statistics, its weekly deaths graph looks wacky.

Finally, looking at the absolute numbers on new cases gives a different list of states, with Florida, Texas, California, Georgia, and North Carolina leading the nation.

Note: imported from Covid Proxima. Click here to see all the interactive elements.