: New:

I continue to check my little dashboard of San Francisco COVID-19 numbers each day to decide whether doing inessential activities in public is foolhardy or just good fun. Lately, all the numbers have been below the pretty-safe line, yay. (Annoyingly, the test positivity % is just barely below the pretty-safe line. If I'd gotten around to blogging about this yesterday, that graph would look much more reassuring.)
line graph charting three values: new case count, test positivity, wastewater levels

You might recall that when California ended the COVID-19 health emergency and thus stopped forcing so many healthcare places to test so many workers+patients, SF's number of reported cases went down, but the % of tests with COVID stayed high. Back in the emergency lots-of-testing days, those numbers followed each other. But in the new less-testing world, they didn't.

Starting in mid-May, the % of tests with COVID dropped below the pretty-safe line, yay. Annoyingly, the %-line and the cases-line still aren't really following each other. On a few days in mid-May, 100s more people than usual got tested; despite that, there weren't more new cases detected than other days. On the one hand, it's reassuring; it suggests I wasn't crazy to do nonessential things despite that %-line staying high. On the other hand, it's still spooky how those formerly-correlated numbers now move around pretty independently.

More recently, you might recall that The California Open Data Portal paused sharing wastewater COVID data, so I fell back to using the California Department of Public Health numbers. It took a couple of weeks longer than predicted to un-pause, but the Open Data Portal is indeed posting data again (as seen at the SF Chronicle's new SF Bay Area Wastewater COVID tracker page). Thus, I resumed using the open data portal's "raw" data instead of using CDPH's pre-computed numbers. This was nice, since it let me take a lot of "noise" out of the numbers. (With the raw data, I can give more "weight" to large samples than to small samples; but with CDPH's pre-computed numbers, I don't know those sample sizes.) Despite the annoyance of the pause, I'm glad that the Open Data Portal folks paused to clean up their data a bit. It was kind of strange that different fields of their data were called FACILITY NAME and hum_frac_mic_conc. Like, obviously two different people named those two different fields without talking to each other. Now the old FACILITY NAME is called facility_name and I get the feeling that's just the tip of an iceberg of data-cleanup that someone's wanted to take care of for years, now finally dealt with.

Tags: programming

lahosken@gmail.com

Tags