OK. I have 20 minutes to kill and another small Amateur Adventure in Covid Stats.
Today: since data for any one day's infections is reported gradually over the next week, what can we learn about the daily infections early on?
So on any day, the reporting looks like this:
Today: since data for any one day's infections is reported gradually over the next week, what can we learn about the daily infections early on?
So on any day, the reporting looks like this:
Does that mean infections are falling? NO - because most of the data for 3rd-8th October is still to come in.
How much is left to come?
Well, I have the past five days' spreadsheets to examine...
Which look like this, compiled:
How much is left to come?
Well, I have the past five days' spreadsheets to examine...
Which look like this, compiled:
So you see on the first day, nothing. Then a lot in the second day, more in the third, then it diminishes. How much by?
Well, I sorted it out, put it in a table like this
Well, I sorted it out, put it in a table like this
Now that *looks* like you get on average 5300 on the first day, 10,000 on the last. But this series is rapidly changing i.e. growing. So better to express each day as a % of the final maximum, like this
This suggests, from the two days for which I have a long enough series (2nd and 3rd October) that on your first big reporting day (day 2) you tend to score around 35% of your eventual total. By day 4 you have 90% of your readings in.
If so, the dates of 5th and 6th October, by recording 6-7,000 on their first big day (7th and 8th), will eventually end up being close to 20,000 infections.
If so, the reasoning in this earlier thread suggests it is worrying https://twitter.com/Gilesyb/status/1313536968639291392
If so, the reasoning in this earlier thread suggests it is worrying https://twitter.com/Gilesyb/status/1313536968639291392
Of course, I could be wrong: it could be that they have suddenly got quicker at gathering the data in, and so much more of the eventual total for those days is already compiled.
Let's see! Public predictions enable me to learn ...
Let's see! Public predictions enable me to learn ...
As Chris G has pointed out, Chris D has done this over a longer time frame, showing it to wax and wane terribly... https://twitter.com/chrisdrakeuk/status/1314231492176994304?s=20