COVID-19

Covid-19 discussion, bring your own statistics
Post Reply
User avatar
shpalman
Stummy Beige
Posts: 3611
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: COVID-19

Post by shpalman » Sun Jan 17, 2021 8:03 am

jimbob wrote:
Sun Jan 17, 2021 1:17 am
Ivor Cummins seems to like my graph. But he's even more of an innumerate idiot than I thought

https://twitter.com/FatEmperor/status/1 ... 75136?s=20

Image

Image
The area under the curve could be interpreted as something like "number of person-days lost" as compared to the baseline.

But
Sweden-total-deaths.png
Sweden-total-deaths.png (96.71 KiB) Viewed 741 times
is already the integral of
Image

The issue is that by subtracting the baseline "number of deaths which would normally have happened by that day" in your graph you've made it seem like you can treat it like the deaths-per-day graph, and integrate it to get the total deaths, which you can't because that's not what it is.
molto tricky

User avatar
jimbob
After Pie
Posts: 2206
Joined: Mon Nov 11, 2019 4:04 pm
Location: High Peak/Manchester

Re: COVID-19

Post by jimbob » Sun Jan 17, 2021 11:55 am

shpalman wrote:
Sun Jan 17, 2021 8:03 am
jimbob wrote:
Sun Jan 17, 2021 1:17 am
Ivor Cummins seems to like my graph. But he's even more of an innumerate idiot than I thought

https://twitter.com/FatEmperor/status/1 ... 75136?s=20

Image

Image
The area under the curve could be interpreted as something like "number of person-days lost" as compared to the baseline.

But
Sweden-total-deaths.png
is already the integral of
Image

The issue is that by subtracting the baseline "number of deaths which would normally have happened by that day" in your graph you've made it seem like you can treat it like the deaths-per-day graph, and integrate it to get the total deaths, which you can't because that's not what it is.

I think you are being too generous. Cummins *loves* talking about excess deaths as being the "gold standard" metric. This implies that he claims to understand them.

If you use any measure of an excess, you need to have a baseline from which this excess occurs. You can then sum those numbers in order to get another way of estimating how many unexpected deaths have occurred in total over the year, and, by looking at the other years, an estimation of the usual variation.

Of course with England and Wales, I have every year this century to look at, so can get a far better idea of the variation, rather than the 6-years including 2020 that I can find for Sweden.
Have you considered stupidity as an explanation

User avatar
shpalman
Stummy Beige
Posts: 3611
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: COVID-19

Post by shpalman » Sun Jan 17, 2021 1:57 pm

Not yet sure how to get weekly data out of ISTAT but here's an interactive thing for visualizing monthly deaths 2020 vs. 2015-2019 for Italy in which you can filter in various ways.

This is filtered to just be Lombardy.
Lombardy-monthly.png
Lombardy-monthly.png (57.46 KiB) Viewed 679 times
I seriously don't understand what insight you think you'd gain by shifting the curves in the lower panel so the blue one sat at zero, but you would lose the sense of the "normal" death rate to compare the magnitude of the 2020 excess to.
molto tricky

User avatar
shpalman
Stummy Beige
Posts: 3611
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: COVID-19

Post by shpalman » Sun Jan 17, 2021 3:29 pm

Thanks to Lincolnshire Live for going behind the Sunday Times's paywall and telling us what's there:

The Sunday Times said a “Cabinet deal” has been done to approve a three-point plan for lifting coronavirus restrictions from early March.

Who have they done the deal with? The virus?
molto tricky

User avatar
headshot
Snowbonk
Posts: 509
Joined: Tue Nov 12, 2019 9:40 am

Re: COVID-19

Post by headshot » Sun Jan 17, 2021 4:38 pm

shpalman wrote:
Sun Jan 17, 2021 3:29 pm
Thanks to Lincolnshire Live for going behind the Sunday Times's paywall and telling us what's there:

The Sunday Times said a “Cabinet deal” has been done to approve a three-point plan for lifting coronavirus restrictions from early March.

Who have they done the deal with? The virus?
The barely moderate MPs in the Tory party have done a deal with the absolutely f.cking lunatics...

User avatar
JQH
Dorkwood
Posts: 1260
Joined: Mon Nov 11, 2019 3:30 pm
Location: Sar Flandan

Re: COVID-19

Post by JQH » Sun Jan 17, 2021 4:43 pm

shpalman wrote:
Sun Jan 17, 2021 3:29 pm
Thanks to Lincolnshire Live for going behind the Sunday Times's paywall and telling us what's there:

The Sunday Times said a “Cabinet deal” has been done to approve a three-point plan for lifting coronavirus restrictions from early March.

Who have they done the deal with? The virus?
This week the Prime Minister said there were signs the lockdown was beginning to work
SO WHY STOP DOING IT YOU f.ck-WITTED PUBLIC SCHOOL TOSS-POT???

And breathe ...
And remember that if you botch the exit, the carnival of reaction may be coming to a town near you.

Fintan O'Toole

User avatar
jimbob
After Pie
Posts: 2206
Joined: Mon Nov 11, 2019 4:04 pm
Location: High Peak/Manchester

Re: COVID-19

Post by jimbob » Sun Jan 17, 2021 6:44 pm

shpalman wrote:
Sun Jan 17, 2021 1:57 pm
I seriously don't understand what insight you think you'd gain by shifting the curves in the lower panel so the blue one sat at zero, but you would lose the sense of the "normal" death rate to compare the magnitude of the 2020 excess to.
It depends what you are doing.

If you are looking at the percentage effect, then no. If you are doing an honest analysis, you can look at the overall year and the slopes, and that works too.

If you are responding to a mendacious fad-diet guru, who claims that this winter is not much worse than 2018, and that COVID-19 is just flu, then it's useful to look at the usual range in total deaths for the time period concerned.

If the previous 5 years had a variation of 2000 over the time period you are looking at - roughly from March through December, then it's reasonable to point out that the 7000 above the average (and possibly more given the time lag in reportiong) is well outside normal variation. And more outside the variation than you'd think if you included the three months of 2020 when COVID hadn't yet hit Sweden.
Have you considered stupidity as an explanation

User avatar
shpalman
Stummy Beige
Posts: 3611
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: COVID-19

Post by shpalman » Mon Jan 18, 2021 1:31 pm

James Annan reckons that (a) the lockdown is working and (b) the decision to not cancel Christmas may end up costing about 30,000 additional deaths.
It seems plausible that an large part of the reason for the striking success of the suppression is that the transmission of the new variant was predominantly enhanced in the young and therefore closing schools has had a particularly strong effect.
molto tricky

User avatar
shpalman
Stummy Beige
Posts: 3611
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: COVID-19

Post by shpalman » Mon Jan 18, 2021 2:37 pm

Clamping down on lockdown rule breakers will get coronavirus back under control, Priti Patel said

The thing is, they probably are under control now, that the lockdown is starting to take effect.

If they weren't under control until now it's directly as a result of government policy to e.g. not close schools and then allow some limited mixing at Christmas.

And if R is "currently estimated to be around 1.2 to 1.3" well, that may be the "current" estimate, but it's not based on the "current" data in which the case rate is quite obviously coming down.
molto tricky

User avatar
lpm
Stummy Beige
Posts: 2787
Joined: Mon Nov 11, 2019 1:05 pm

Re: COVID-19

Post by lpm » Mon Jan 18, 2021 2:43 pm

What I think James Annan (though probably more other commentators) might be missing is that the lockdown really began on 26 Dec, even though the official date was 5 Jan. The stats for mobility etc show 8-10 days of very low activity, followed by a step up with the return to work on Mon 4 Jan.

So I think the lockdown looks to be working particularly well in its first two weeks - reflecting activities 3 weeks ago - but then the figures might have a slower improvement over the next couple of weeks.

If we can hold the govt to this lockdown to end of Feb, and then only a sliver of unlockdown in March, then UK cases should be in very good shape by April. They'll f.ck it up though.
What ever happened to that Trump guy, you know, the one who was president for a bit?

KAJ
Clardic Fug
Posts: 197
Joined: Thu Nov 14, 2019 5:05 pm
Location: UK

Re: COVID-19

Post by KAJ » Mon Jan 18, 2021 5:46 pm

lpm wrote:
Mon Jan 18, 2021 2:43 pm
What I think James Annan (though probably more other commentators) might be missing is that the lockdown really began on 26 Dec, even though the official date was 5 Jan. The stats for mobility etc show 8-10 days of very low activity, followed by a step up with the return to work on Mon 4 Jan.
<snip>
Yes. This is my log-quadratic fit, zero-weighting the latest 5 (marked as Incomplete) and Xmas day and NY day.
SpecCases.png
SpecCases.png (13.73 KiB) Viewed 465 times
The fit isn't as good as it used to be, but I think the slowdown before 5 January is clear.

Code: Select all

Analysis of Variance Table

Response: log(SpecCases)
              Df  Sum Sq Mean Sq F value    Pr(>F)    
poly(date, 2)  2 1.05647 0.52824 17.3170 7.922e-05 ***
day            6 0.50266 0.08378  2.7464   0.04714 *  
Residuals     17 0.51857 0.03050                      
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Coefficients:
               Estimate Std. Error t value Pr(>|t|)    
(Intercept)    10.58169    0.08998 117.596  < 2e-16 ***
poly(date, 2)1  0.19941    0.28378   0.703 0.491773    
poly(date, 2)2 -1.16537    0.27875  -4.181 0.000627 ***
dayMon          0.26508    0.12358   2.145 0.046702 *  
dayTue          0.32283    0.12388   2.606 0.018451 *  
dayWed          0.24866    0.12476   1.993 0.062546 .  
dayThu          0.04356    0.12417   0.351 0.730034    
dayFri          0.13232    0.15432   0.857 0.403107    
daySat         -0.05186    0.12357  -0.420 0.679990    
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.1747 on 17 degrees of freedom
Multiple R-squared:  0.7504	

User avatar
sTeamTraen
Dorkwood
Posts: 1583
Joined: Mon Nov 11, 2019 4:24 pm
Location: Palma de Mallorca, Spain

Re: COVID-19

Post by sTeamTraen » Mon Jan 18, 2021 6:52 pm

Deaths recorded on the last two days (Sunday and Monday, always the quietest on this front) are 16% up on last weekend, but cases have clearly turned the corner and deaths are about 8-10 days behind. On that basis I'm guessing an average of 1,500 deaths per day for the next 5 days and then a slight fall the following week, for a total of close to 30,000 deaths in January.

30 f.cking thousand. :shock:

4cb61bd9-06d5-4226-bd3b-622c39e7bf6c.png
4cb61bd9-06d5-4226-bd3b-622c39e7bf6c.png (9.6 KiB) Viewed 435 times
Something something hammer something something nail

User avatar
headshot
Snowbonk
Posts: 509
Joined: Tue Nov 12, 2019 9:40 am

Re: COVID-19

Post by headshot » Mon Jan 18, 2021 6:57 pm

I hope people enjoyed their turkey and mince pies and feel it was all worth it. Dickheads.

User avatar
Bird on a Fire
Light of Blast
Posts: 5241
Joined: Fri Oct 11, 2019 5:05 pm
Location: nadir of brie

Re: COVID-19

Post by Bird on a Fire » Mon Jan 18, 2021 6:57 pm

Portugal figures.

After barely having a first wave, thanks to a strict lockdown and travel restrictions, we then relaxed over the summer until things started getting bad again. Got more lockdowny again in October and again in November - but with a couple of temporary relaxations over Christmas and New Years, the effects of which are still very clear to see on this chart, and the numbers haven't started going down again - full government-ordered lockdown only started again on 15th January, so will take a while I expect.
Screenshot_2021-01-18_18-46-28.png
Screenshot_2021-01-18_18-46-28.png (40.61 KiB) Viewed 428 times
My municipality is still orange (highly elevated risk: 480-960 cases per 100k; we're at 917, compared with 550 last week) but most of the map is red now. Total deaths have passed the 9000 mark (out of 10 million population), with around 150-200 a day at the moment.

So it's pretty sh.t really. Not UK sh.t, but sh.t nonetheless.

I think it goes to show that you really can't afford to relax, even for a 48 hour period.
He has the grace of a swan, the wisdom of an owl, and the eye of an eagle—ladies and gentlemen, this man is for the birds!

User avatar
Bird on a Fire
Light of Blast
Posts: 5241
Joined: Fri Oct 11, 2019 5:05 pm
Location: nadir of brie

Re: COVID-19

Post by Bird on a Fire » Mon Jan 18, 2021 7:02 pm

Forgot to mention - the only main exception from the lockdown is in-person voting for the presidential elections.

They've split it over two weekends instead of the usual one, masks-and-distancing as standard when outside one's house, and within the polling stations they have extra tables and stuff. But still, given that they knew this was coming the whole f.cking time they could have arranged postal voting and moved Portugal into the 21st century. Turnout is normally only about 50% and I can't imagine a f.cking pandemic is going to help, even with far-right populist/c.nt Andre Ventura lurking menacingly in the polls.
He has the grace of a swan, the wisdom of an owl, and the eye of an eagle—ladies and gentlemen, this man is for the birds!

User avatar
lpm
Stummy Beige
Posts: 2787
Joined: Mon Nov 11, 2019 1:05 pm

Re: COVID-19

Post by lpm » Tue Jan 19, 2021 11:56 am

UK: December

1/10 antibodies
1/1,000 dead
IFR = 1%

There's been nothing particularly surprising in our Covid journey. We pretty much knew in March there'd be loads of cases, probably multiple waves, around 1% IFR, and probably a vaccine at some point in 2021. Have we made modelling and predicting stuff and planning seem harder than it than it actually was?
What ever happened to that Trump guy, you know, the one who was president for a bit?

User avatar
shpalman
Stummy Beige
Posts: 3611
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: COVID-19

Post by shpalman » Tue Jan 19, 2021 12:07 pm

lpm wrote:
Tue Jan 19, 2021 11:56 am
UK: December

1/10 antibodies
1/1,000 dead
IFR = 1%

There's been nothing particularly surprising in our Covid journey. We pretty much knew in March there'd be loads of cases, probably multiple waves, around 1% IFR, and probably a vaccine at some point in 2021. Have we made modelling and predicting stuff and planning seem harder than it than it actually was?
Any idiot can order a full lockdown.
molto tricky

User avatar
lpm
Stummy Beige
Posts: 2787
Joined: Mon Nov 11, 2019 1:05 pm

Re: COVID-19

Post by lpm » Tue Jan 19, 2021 12:19 pm

Any idiot can shut a stable door.

Our idiot struggled with that task, however, let alone the timing of the task.
What ever happened to that Trump guy, you know, the one who was president for a bit?

User avatar
shpalman
Stummy Beige
Posts: 3611
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: COVID-19

Post by shpalman » Tue Jan 19, 2021 4:32 pm

Any idiot can see whether the trend in cases by date reported over the past week or so was numbers going up, and order a lockdown.

It takes real intelligence to have to wait 4-5 days for all the "by specimen date" data to finish coming through and then spend a week analysing it to get an R number and then spend a week modelling the effects of various restrictions and then another few days hinting that new restrictions are coming in before ordering some sort of half-arsed and unenforced lockdown.
molto tricky

User avatar
shpalman
Stummy Beige
Posts: 3611
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: COVID-19

Post by shpalman » Tue Jan 19, 2021 4:54 pm

UK's number of deaths today is particularly high because of catching up with yesterday and the day before which were obviously low, the Guardian doesn't say.

This is especially annoying because their standard template for reporting the figures from Italy, say, is "today's number of cases is N which lower/higher than yesterday's number which was Nd-1" as if the daily variation tells you anything other than what day of the week it is.
molto tricky

KAJ
Clardic Fug
Posts: 197
Joined: Thu Nov 14, 2019 5:05 pm
Location: UK

Re: COVID-19

Post by KAJ » Tue Jan 19, 2021 5:14 pm

Hmm. Sunday and Monday reported deaths are systematically lower than later days of the week, but today's high number isn't entirely a weekday effect. As the Grauniad says:
This is the largest daily figure for Covid deaths (defined as deaths within 28 days of testing positive) recorded so far on a single day.

The previous biggest high came on Wednesday last week, when 1,564 deaths were recorded.
... so the reported number today is higher than any previous Tuesday. Last Wednesday's was higher than any previous Wednesday.
PubDeaths.png
PubDeaths.png (16.31 KiB) Viewed 266 times

Code: Select all

At 19/01 fit = 1408.3 with doubling time = 19.6 days. That time doubling in 44.1 days 

Analysis of Variance Table

Response: log(PubDeaths)
              Df Sum Sq Mean Sq F value    Pr(>F)    
poly(date, 2)  2 2.5829 1.29146  25.321 4.374e-06 ***
day            6 3.8510 0.64183  12.584 9.727e-06 ***
Residuals     19 0.9691 0.05100                      
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Call:
lm(formula = as.formula(aform), data = aDF, weights = weights)

Residuals:
     Min       1Q   Median       3Q      Max 
-0.47246 -0.13848  0.02321  0.13528  0.35104 

Coefficients:
               Estimate Std. Error t value Pr(>|t|)    
(Intercept)     6.13330    0.11307  54.243  < 2e-16 ***
poly(date, 2)1  1.95783    0.23310   8.399 8.07e-08 ***
poly(date, 2)2 -0.12368    0.22624  -0.547 0.590948    
dayMon         -0.08671    0.15980  -0.543 0.593691    
dayTue          0.54543    0.16017   3.405 0.002969 ** 
dayWed          0.95700    0.16128   5.934 1.03e-05 ***
dayThu          0.81578    0.16054   5.081 6.63e-05 ***
dayFri          0.72337    0.16006   4.519 0.000235 ***
daySat          0.27624    0.15979   1.729 0.100064    
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.2258 on 19 degrees of freedom
Multiple R-squared:  0.8691

User avatar
lpm
Stummy Beige
Posts: 2787
Joined: Mon Nov 11, 2019 1:05 pm

Re: COVID-19

Post by lpm » Tue Jan 19, 2021 5:20 pm

The average for the last seven days is 1,181 deaths a day. This is world-beating, even without a one day spike.
What ever happened to that Trump guy, you know, the one who was president for a bit?

User avatar
sTeamTraen
Dorkwood
Posts: 1583
Joined: Mon Nov 11, 2019 4:24 pm
Location: Palma de Mallorca, Spain

Re: COVID-19

Post by sTeamTraen » Tue Jan 19, 2021 5:49 pm

Opti wrote:
Sat Jan 02, 2021 6:20 pm
There we are then. Gibraltar is f.cked. Incredibly fast acceleration of cases. All the adjoining ayuntamientos in Cadiz province are now locked down as well.
Covid don't give a sh.t about anyone taking control of are borders.
Only a day or so ago, you could fly from UK to Gib without a prior PCR test. Tested only on arrival. By then, you were Gib's problem.
47 deaths now in Gibraltar, which has the same population (~34,000) as Billericay, Morecambe, or Exmouth. :cry:
Something something hammer something something nail

User avatar
sTeamTraen
Dorkwood
Posts: 1583
Joined: Mon Nov 11, 2019 4:24 pm
Location: Palma de Mallorca, Spain

Re: COVID-19

Post by sTeamTraen » Tue Jan 19, 2021 5:52 pm

lpm wrote:
Tue Jan 19, 2021 5:20 pm
The average for the last seven days is 1,181 deaths a day. This is world-beating, even without a one day spike.
I'm guessing deaths will peak this week; that is, the total reported on the seven days from 17-23 January will be slightly higher than the total reported from 24-30 January. My best guess at the total deaths reported in January is 30,000, with the UK all-time total hitting 100,000 around 29 January.
Something something hammer something something nail

User avatar
jimbob
After Pie
Posts: 2206
Joined: Mon Nov 11, 2019 4:04 pm
Location: High Peak/Manchester

Re: COVID-19

Post by jimbob » Tue Jan 19, 2021 5:58 pm

lpm wrote:
Tue Jan 19, 2021 12:19 pm
Any idiot can shut a stable door.

Our idiot struggled with that task, however, let alone the timing of the task.
Three times
Have you considered stupidity as an explanation

Post Reply