COVID-19

Covid-19 discussion, bring your own statistics
Post Reply
User avatar
lpm
Stummy Beige
Posts: 2778
Joined: Mon Nov 11, 2019 1:05 pm

Re: COVID-19

Post by lpm » Wed Dec 23, 2020 4:41 pm

shpalman wrote:
Wed Dec 23, 2020 4:32 pm
Congratulations on your 39237 new covids.

How were the scotch eggs?
Well worth it. I predict from now on the English will get pissed in the pub then get a scotch egg instead of a kebab on the way home.
What ever happened to that Trump guy, you know, the one who was president for a bit?

User avatar
sTeamTraen
Dorkwood
Posts: 1579
Joined: Mon Nov 11, 2019 4:24 pm
Location: Palma de Mallorca, Spain

Re: COVID-19

Post by sTeamTraen » Wed Dec 23, 2020 7:06 pm

Anyone want to take a wild guess at what might happen to the red line over the next month? :shock:

Untitled.png
Untitled.png (19.95 KiB) Viewed 635 times
Something something hammer something something nail

User avatar
shpalman
Stummy Beige
Posts: 3590
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: COVID-19

Post by shpalman » Wed Dec 23, 2020 7:16 pm

Deaths by date reported are going up, but deaths by date of death are remaining constant, as we were discussing on the previous page of this thread here earlier before.

That's weird but I suppose it means that there's catching up to do with registering the deaths from the whole of the past month or so. Maybe if KAJ has every day's "deaths by date of death" data archived we'd be able to see.
molto tricky

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

Re: COVID-19

Post by KAJ » Wed Dec 23, 2020 7:45 pm

shpalman wrote:
Wed Dec 23, 2020 7:16 pm
Deaths by date reported are going up, but deaths by date of death are remaining constant, as we were discussing on the previous page of this thread here earlier before.

That's weird but I suppose it means that there's catching up to do with registering the deaths from the whole of the past month or so. Maybe if KAJ has every day's "deaths by date of death" data archived we'd be able to see.
I've got the data back to 7 November. I have a work-in-progress R Markdown notebook which I run each day. Today's run produced this (inter alia):

Code: Select all

getcData("cDeaths.rds", maxlag = 20, mincount = 20)$plot + scale_y_continuous(trans = "exp")
I restrict to 'Dates of Death' with 20 or more reporting dates to ensure (as far as reasonable) that the maximum value for that date is the final value.
cdeaths.png
cdeaths.png (29.02 KiB) Viewed 618 times
.

It's clear to me that the 'incomplete' dates extend far further than the 5 days so marked by coronavirus.data.gov.uk. I wouldn't rely on a date being substantially complete until 12 days after.

User avatar
shpalman
Stummy Beige
Posts: 3590
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: COVID-19

Post by shpalman » Wed Dec 23, 2020 8:02 pm

I've only been saving the cases count from Lincolnshire, which is always by date of test. (I may have missed a few days).
Lincolnshire-case-development.png
Lincolnshire-case-development.png (103.58 KiB) Viewed 609 times
Of course we know that most cases have been reported after 4-5 days so each day's data has a quick fall-off at the end which you see gets filled in from the next few traces.

I was hoping to see a similar graph for the UK's deaths by date of death in which we'd see the whole curve of November-December moving upwards, or at least the post-peak decrease slowly becoming less of a decrease. Or at least comparing the last few days' worth of reports to see where all these more-than-average deaths which are being reported date from.
molto tricky

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

Re: COVID-19

Post by KAJ » Wed Dec 23, 2020 8:04 pm

shpalman wrote:
Wed Dec 23, 2020 7:16 pm
Deaths by date reported are going up, but deaths by date of death are remaining constant, as we were discussing on the previous page of this thread here earlier before.

That's weird but I suppose it means that there's catching up to do with registering the deaths from the whole of the past month or so. Maybe if KAJ has every day's "deaths by date of death" data archived we'd be able to see.
Based on the previous 33 dates, zero-weighting the latest 5, and fitting quadratics (which really don't fit very well!) yesterday looked like a slow decline:
download.png
download.png (28.3 KiB) Viewed 605 times

Code: Select all

Coefficients:
##                Estimate Std. Error t value Pr(>|t|)    
## (Intercept)     6.05090    0.01410 429.278  < 2e-16 ***
## poly(date, 2)1 -0.29870    0.09847  -3.033  0.00557 ** 
## poly(date, 2)2  0.01786    0.09544   0.187  0.85310    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.06323 on 25 degrees of freedom
## Multiple R-squared:  0.3692, Adjusted R-squared:  0.3188 
Today looks flat
Deaths.png
Deaths.png (9.58 KiB) Viewed 605 times

Code: Select all

Coefficients:
               Estimate Std. Error t value Pr(>|t|)    
(Intercept)     6.07828    0.01456 417.565   <2e-16 ***
poly(date, 2)1 -0.06750    0.10169  -0.664    0.513    
poly(date, 2)2  0.16405    0.09856   1.664    0.109    
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.0653 on 25 degrees of freedom
Multiple R-squared:  0.2093,	Adjusted R-squared:  0.1461 
F-statistic:  3.31 on 2 and 25 DF,  p-value: 0.05307
I reckon the higher numbers are working through and I'll start seeing upward slopes in the next day or two.

User avatar
jimbob
After Pie
Posts: 2199
Joined: Mon Nov 11, 2019 4:04 pm
Location: High Peak/Manchester

Re: COVID-19

Post by jimbob » Wed Dec 23, 2020 8:14 pm

sTeamTraen wrote:
Tue Dec 22, 2020 10:57 pm
shpalman wrote:
Tue Dec 22, 2020 9:56 pm
The ONS stats don't suggest that there's a significant number of excess but not-officially-covid deaths on top of the significant number of excess covid deaths in the second wave.
I've done something similar with data from the Netherlands (the five-year average deaths are standardized to zero). In wave 2, most of the excess deaths are COVID-positive. In week 33 there was a heatwave which killed a few older people. But you can see that back in March-April a lot of people were very probably dying from COVID without getting a test.

Deaths in the 0-65 age range are very slightly below normal, although probably within fluctuation range. I guess this corresponds to some reduction from road traffic deaths and other accidents outside the home, offset by some COVID deaths in the upper third of that age range.

Untitled.png
Lots of suppression of other infectious diseases as well. And not just the obvious respiratory diseases. Also even food poisoning.
Have you considered stupidity as an explanation

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

Re: COVID-19

Post by KAJ » Wed Dec 23, 2020 8:24 pm

shpalman wrote:
Wed Dec 23, 2020 8:02 pm
I've only been saving the cases count from Lincolnshire, which is always by date of test. (I may have missed a few days).

Lincolnshire-case-development.png

Of course we know that most cases have been reported after 4-5 days so each day's data has a quick fall-off at the end which you see gets filled in from the next few traces.

I was hoping to see a similar graph for the UK's deaths by date of death in which we'd see the whole curve of November-December moving upwards, or at least the post-peak decrease slowly becoming less of a decrease. Or at least comparing the last few days' worth of reports to see where all these more-than-average deaths which are being reported date from.
I'm not sure, but I think this may be what you wanted
Deaths2.png
Deaths2.png (62.51 KiB) Viewed 595 times
(ggplot really is nice, this is the code that produced that graph)

Code: Select all

ggplot(data = cDeathsDF, aes(x = Published, y = DateDeaths, colour = as.factor(date))) + geom_line( ) 

User avatar
shpalman
Stummy Beige
Posts: 3590
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: COVID-19

Post by shpalman » Wed Dec 23, 2020 9:22 pm

Not sure, the curves look like they're the wrong way around.

What I want is deaths-per-day versus date-of-death for each data set reported on a different day.
molto tricky

plodder
Dorkwood
Posts: 1272
Joined: Mon Nov 11, 2019 1:50 pm

Re: COVID-19

Post by plodder » Thu Dec 24, 2020 2:34 pm

lpm wrote:
Fri Apr 17, 2020 9:41 am
... which is what Nightingales should have been used for?

- Build a non-Covid Nightingale for people recovering from normal illness and not yet well enough to go back to care homes, and do everything possible to test and keep it Covid-free

- Build a separate Covid Nightingale for recoverers - survivors discharged from ICU, spend a day or two in regular hospital wards, then moved to a specialist Nightingale for however many days it takes to be well enough to go home. Johnson was lucky, in that he could leave hospital promptly to recover in a second home in the country, waited on by servants, but many people need a couple of weeks in hospital or more

Much of this is the persistent NHS problem of how to care for elderly and frail people, not ill enough for a full hospital bed, not well enough to return to independent life at home or a care home. Covid is a multiplier of this - 85% of over-80s survive, but a fairly large proportion of these will be severely knocked back and in need of extended after-care. The headline death rate is only part of the problem - the additional care burden is far larger and if not addressed it will lead to premature deaths months later among patients not recovering their fitness/mobility.
It’s gone.

https://mobile.twitter.com/TiceRichard/ ... 2773159936

User avatar
bob sterman
Fuzzable
Posts: 366
Joined: Mon Nov 11, 2019 10:25 pm

Re: COVID-19

Post by bob sterman » Thu Dec 24, 2020 2:46 pm

plodder wrote:
Thu Dec 24, 2020 2:34 pm
It’s gone.

https://mobile.twitter.com/TiceRichard/ ... 2773159936
Maybe it has - but I wouldn't believe a word that guy Richard Tice says. He's the Brexit Party chair - known as "Farage's sidekick" and is one of these folk who thinks most cases are PCR false positives.

Obviously well qualified to comment on such issues as a property investor.

His idea of a good public health information campaign is the Wetherspoon's newsletter...

https://mobile.twitter.com/TiceRichard/ ... 1141277697

plodder
Dorkwood
Posts: 1272
Joined: Mon Nov 11, 2019 1:50 pm

Re: COVID-19

Post by plodder » Thu Dec 24, 2020 3:49 pm

:shock:

oops. caveats apply etc

OffTheRock
Stargoon
Posts: 91
Joined: Mon Dec 02, 2019 6:52 pm

Re: COVID-19

Post by OffTheRock » Thu Dec 24, 2020 4:26 pm

As far as I’m aware it was mothballed but can be stood up again really quickly. Plan was for it only to take patients that are already ventilated so I doubt they’d reopen it unless they run out of critical care space in London & SE.

User avatar
sTeamTraen
Dorkwood
Posts: 1579
Joined: Mon Nov 11, 2019 4:24 pm
Location: Palma de Mallorca, Spain

Re: COVID-19

Post by sTeamTraen » Thu Dec 24, 2020 4:31 pm

bob sterman wrote:
Thu Dec 24, 2020 2:46 pm
plodder wrote:
Thu Dec 24, 2020 2:34 pm
It’s gone.

https://mobile.twitter.com/TiceRichard/ ... 2773159936
Maybe it has - but I wouldn't believe a word that guy Richard Tice says. He's the Brexit Party chair - known as "Farage's sidekick" and is one of these folk who thinks most cases are PCR false positives.
I think the point here is that the government's handling of this has been so incompetent that even a conspiracist tw.t like Tice can make hay and a lto of sensible people will be nodding along. Apart from one brief moment in that clip where he lets slips that the real point of his story is to show that there is no problem in London ("because they've taken the Nightingale down, so that proves it") and so everyone should get back in the pub immediately, it could equally well have been made by a principled opposition party to show that the whole idea of the Nightingales was flawed from day one because there are no doctors or nurses to staff them. For example, Tice has fooled the otherwise admirable Gary Neville into thinking it was about resources.

Indeed, the same question could have been asked back when the government was about to order 30,000 ventilators or however many it was; it turns out that you can't just get a bloke from Kwik-Fit, or even a regular nurse, to just stand by it and press button B when it goes "beep".
Something something hammer something something nail

User avatar
shpalman
Stummy Beige
Posts: 3590
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: COVID-19

Post by shpalman » Thu Dec 24, 2020 4:40 pm

It also turns out that you don't really need that many of that sort of ventilator and can help a lot more people with a much simpler CPAP-style thing.
molto tricky

User avatar
shpalman
Stummy Beige
Posts: 3590
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: COVID-19

Post by shpalman » Thu Dec 24, 2020 5:27 pm

shpalman wrote:
Wed Dec 23, 2020 4:32 pm
Congratulations on your 39237 new covids.

How were the scotch eggs?
Showing that yesterday was no fluke, we have 39036 new covids today.

The UK is increasing with a 10-11 day doubling time.
molto tricky

User avatar
shpalman
Stummy Beige
Posts: 3590
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: COVID-19

Post by shpalman » Thu Dec 24, 2020 5:46 pm

shpalman wrote:
Thu Dec 24, 2020 5:27 pm
shpalman wrote:
Wed Dec 23, 2020 4:32 pm
Congratulations on your 39237 new covids.

How were the scotch eggs?
Showing that yesterday was no fluke, we have 39036 new covids today.

The UK is increasing with a 10-11 day doubling time.
Northern Ireland haven't registered any cases today so there's a few hundred missing from that number and that would make it higher than yesterday's.
molto tricky

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

Re: COVID-19

Post by KAJ » Thu Dec 24, 2020 5:47 pm

shpalman wrote:
Thu Dec 24, 2020 5:27 pm
shpalman wrote:
Wed Dec 23, 2020 4:32 pm
Congratulations on your 39237 new covids.

How were the scotch eggs?
Showing that yesterday was no fluke, we have 39036 new covids today.

The UK is increasing with a 10-11 day doubling time.
Looking at cases by specimen date over 28 days (+ 5 zero weighted) and fitting a quadratic (R-sq = 99%) I find:

Code: Select all

At 22/12 fit = 68505.1 with doubling time = 6.0 days. That time halving in 12.1 days
SpecCases.png
SpecCases.png (12 KiB) Viewed 429 times

Code: Select all

Coefficients:
               Estimate Std. Error t value Pr(>|t|)    
(Intercept)     9.72789    0.02128 457.153  < 2e-16 ***
poly(date, 2)1  2.28040    0.06522  34.965  < 2e-16 ***
poly(date, 2)2  1.10736    0.06194  17.879 2.42e-13 ***
dayMon          0.44647    0.02898  15.404 3.44e-12 ***
dayTue          0.34619    0.02903  11.924 2.89e-10 ***
dayWed          0.32844    0.02912  11.279 7.33e-10 ***
dayThu          0.24560    0.02925   8.395 8.13e-08 ***
dayFri          0.26261    0.02905   9.039 2.61e-08 ***
daySat         -0.00646    0.02899  -0.223    0.826    
---
Signif. codes:  
0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.04096 on 19 degrees of freedom
Multiple R-squared:  0.9898,	Adjusted R-squared:  0.9856

User avatar
discovolante
After Pie
Posts: 2363
Joined: Fri Oct 11, 2019 5:10 pm

Re: COVID-19

Post by discovolante » Thu Dec 24, 2020 5:57 pm

shpalman wrote:
Thu Dec 24, 2020 5:27 pm
shpalman wrote:
Wed Dec 23, 2020 4:32 pm
Congratulations on your 39237 new covids.

How were the scotch eggs?
Showing that yesterday was no fluke, we have 39036 new covids today.

The UK is increasing with a 10-11 day doubling time.
In % of tests terms aren't we roughly back where we were in about mid November?
socialism is when the government does stuff

User avatar
shpalman
Stummy Beige
Posts: 3590
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: COVID-19

Post by shpalman » Thu Dec 24, 2020 6:05 pm

discovolante wrote:
Thu Dec 24, 2020 5:57 pm
shpalman wrote:
Thu Dec 24, 2020 5:27 pm
shpalman wrote:
Wed Dec 23, 2020 4:32 pm
Congratulations on your 39237 new covids.

How were the scotch eggs?
Showing that yesterday was no fluke, we have 39036 new covids today.

The UK is increasing with a 10-11 day doubling time.
In % of tests terms aren't we roughly back where we were in about mid November?
Yes, and it's trending upwards since the whole week has been that high.
molto tricky

User avatar
shpalman
Stummy Beige
Posts: 3590
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: COVID-19

Post by shpalman » Thu Dec 24, 2020 8:12 pm

The numbers over the next few days aren't going to make a whole lot of sense since there'll be days in which the different countries in the country of the UK won't report; only England will report every day but for sure it won't fit the usual weekday dependence.

There might be a bit of a shock as the backlog comes through, of the order of 4-5000 per day for each day of non-reporting.

For there to be an average of 60,000 in the specimen-date data we'd usually need similar numbers in the date-reported data 4-5 days later and since that's when the backlog will be coming through it might peak even higher than that. It seems ridiculous that we might be seeing 80,000 on one of the days next week but maybe we will.
molto tricky

User avatar
sTeamTraen
Dorkwood
Posts: 1579
Joined: Mon Nov 11, 2019 4:24 pm
Location: Palma de Mallorca, Spain

Re: COVID-19

Post by sTeamTraen » Thu Dec 24, 2020 10:24 pm

shpalman wrote:
Thu Dec 24, 2020 4:40 pm
It also turns out that you don't really need that many of that sort of ventilator and can help a lot more people with a much simpler CPAP-style thing.
Yes, that was established quite quickly. Although to be fair, when you know nothing about a disease, the thing that you have experience of that it most closely resembles is your best starting point, and if the UK (or any other) government had said "We don't have any actual evidence that ventilators are what's needed" it would have all been "Heartless Tory scrooges want to kill your Nan".
Something something hammer something something nail

User avatar
sTeamTraen
Dorkwood
Posts: 1579
Joined: Mon Nov 11, 2019 4:24 pm
Location: Palma de Mallorca, Spain

Re: COVID-19

Post by sTeamTraen » Thu Dec 24, 2020 10:27 pm

shpalman wrote:
Thu Dec 24, 2020 8:12 pm
For there to be an average of 60,000 in the specimen-date data we'd usually need similar numbers in the date-reported data 4-5 days later and since that's when the backlog will be coming through it might peak even higher than that. It seems ridiculous that we might be seeing 80,000 on one of the days next week but maybe we will.
Sweden reports on Tuesdays through Fridays, but they closed on Wednesday evening. So their next report will be on 29 December. I'm expecting that to show around 45,000 cases, in a country with a population of 10 million. At the moment they are on the same trajectory as the US, within 5 or 10%, for both cases and deaths. Even the UK isn't doing as badly on both.
Something something hammer something something nail

User avatar
bob sterman
Fuzzable
Posts: 366
Joined: Mon Nov 11, 2019 10:25 pm

Re: COVID-19

Post by bob sterman » Fri Dec 25, 2020 7:11 am

plodder wrote:
Thu Dec 24, 2020 3:49 pm
:shock:

oops. caveats apply etc
The Sun appear to have photos of the Excel centre hall empty.

Maybe the gear is all packed away? My apologies, looks like this Tice guy could have been correct about this. Even a stopped clock....

plodder
Dorkwood
Posts: 1272
Joined: Mon Nov 11, 2019 1:50 pm

Re: COVID-19

Post by plodder » Fri Dec 25, 2020 7:58 am

I suppose all those vets wanted their ventilators back?

Post Reply