Kevbad the Bad
Amiable Bowel Syndrome
I’ve just been told that some tossers are organising a ‘clap for Boris’ event tonight at 8pm. I don’t know if it’s local or national. You have been warned.
Coronavirus in the UK - news, lockdown and discussion
- Thread starter Sasaferrato
- Start date
Mr.Bishie
pickled egg
National, seeing that it’s trending in second place on the twatter.
Numbers
an ting!
I’m certain you’re at least 5 or 6 people.The NHS England data that I have been moaning about not having access to in recent weeks is actually available, yay.
This data will allow me to build a picture of hospital deaths where the date is the actual date of death, rather than the date of reporting. Lag is still there, but unlike the daily overall numbers, the accuracy of past days numbers will improve over time.
Statistics » COVID-19 Daily Deaths
Health and high quality care for all, <br />now and for future generationswww.england.nhs.uk
When time allows, I will try to turn this data into something more instantly readable. How quickly that happens likely depends on how interesting the picture that emerges from my first poke at this turns out to be. Its possibly still a bit early to get the most out of even the March figures, not sure yet.
Obviously this data is only for hospitals so I'll have to merge in ONS data too to get the most complete picture possible, or frankly I could jsut wait a very long time and use the ONS data on its own.
2hats and 6elbows for the win.
planetgeli
There's no future in England's dreaming
FWIW, 854 new deaths reported today.
elbows
Well-Known Member
Well the difference in hospital death data for England is quite pronounced.
The column on the left shows the total number of deaths in hospitals in England that were reported at the time. Because this daily figure is referring to the previous day, I have displaced it by one day to be fairer, eg the final figure of 4897 was reported on March 6th but refers to the period of March 5th so thats where I have put it.
The column on the right shows the new, imporved data, where deaths are actually placed on the day they happened, rather than the day they made it through the reporting system. Unlike the column on the left, the numbers for any particular date are being revised as time goes on. So they eventually give a much more time-accurate picture, but there is still the same lag really. So for example the numbers currently shown for recent dates are likely to increase a lot in future, and even the numbers of deaths from earlier in March can still increase.
So yeah, big differences, even though this is still only hospital deaths. It didnt take very long before lag means the cumulative totals given each day were haf the actual number of hospital deaths that had actually been reached at that point.
Cumulative hospital deaths:
The column on the left shows the total number of deaths in hospitals in England that were reported at the time. Because this daily figure is referring to the previous day, I have displaced it by one day to be fairer, eg the final figure of 4897 was reported on March 6th but refers to the period of March 5th so thats where I have put it.
The column on the right shows the new, imporved data, where deaths are actually placed on the day they happened, rather than the day they made it through the reporting system. Unlike the column on the left, the numbers for any particular date are being revised as time goes on. So they eventually give a much more time-accurate picture, but there is still the same lag really. So for example the numbers currently shown for recent dates are likely to increase a lot in future, and even the numbers of deaths from earlier in March can still increase.
So yeah, big differences, even though this is still only hospital deaths. It didnt take very long before lag means the cumulative totals given each day were haf the actual number of hospital deaths that had actually been reached at that point.
Cumulative hospital deaths:
prunus
Tick tock.
Thank you so much for doing this, it really illustrates just how much garbage in is going into anything I am modelling (using just the announced figures). I think I might just stop now.
I don't know how you have the data organised, but if it's easy to extract, a matrix of how each day's total evolves over time would be really useful... (ie 'death day' as the row headings in column one, with 'estimate for deaths on day' as the row headings - each would start with the announced number on day n+1 and evolve over time until stabilises at the truth. I suspect the data availabilty may not be amenable to such a presentation. but maybe...? Basically I'm asking for a way to model the 'true' deaths from the announced figure I guess! Maybe just 2x, as you say...
editor
hiraethified
Dealers having to adapt
www.vice.com
Drug Dealers Are Dressing Up as Delivery Drivers, Joggers and Nurses to Sell Heroin During Lockdown
We examined the impact of coronavirus and social distancing on Britain’s lucrative street heroin and crack trade.
www.vice.com
andysays
Love and solidarity
Sorry, didn't see you were grinning under your mask
maomao
普費斯
Should do a clap for the virus one at 9.
elbows
Well-Known Member
Thanks. I'm not someone who attempts modelling or clever predictions or even sophisticated presentation of data. I just muck around with the raw data and comparisons to other raw data. I dont really propose 2X as a rule that could be used for modelling, its just the typical sort of observation I was able to make by looking at the raw numbers.
I am very interested in the results of anything other people can do on that front though, I just cant help much other than to point at the most pertinent data sources.
In this case, I manually collected the daily announced England totals over time, and then I would just used the England row in the COVID19 total deaths by region tab of this, keeping in mind that any of the entries from it may change in future:
COVID 19 total announced deaths 7 April 2020 (from Statistics » COVID-19 Daily Deaths )
When I did my above numbers that one for the 7th was not yet available, I will re-do my exercise and see how it compares after just one days changes.
littlebabyjesus
one of Maxwell's demons
That's going to be really tricky, I think. At some point, as we pass peak, the daily figure is going to be more than the actual figure for a while. Is there enough information in there to predict when that might be?
It's a good illustration of how hard it is to judge where we're at. I'm increasingly thinking that the single number showing 'how many people are currently in hospital' is probably the best measure of that.
elbows
Well-Known Member
Yes, I am using hospital data for the best balance of near-realtime clues right now. Plus I look at it on a regional basis so that I do not miss trends happening in particular regions which could distort the overall national picture. Luckily we actually get that data now, via the slides and datasets that are published in connection with the daily number 10 press briefings.
www.gov.uk
Slides, datasets and transcripts to accompany coronavirus press conferences
Slides, datasets and transcripts from press conferences at 10 Downing Street in response to coronavirus.
elbows
Well-Known Member
I completed my update of the previous figures. Added todays published data as a 3rd column so we can see how much the numbers have been corrected compared to yesterday. Note that sometimes there will be a revision downwards, one of the earlier dates had 1 death removed from it compared to yesterday by the looks of it.
By the way, the raw data from the NHS shows number of deaths per day, rather than the cumulative totals I always seem to go for when I reformat & share data here.
By the way, the raw data from the NHS shows number of deaths per day, rather than the cumulative totals I always seem to go for when I reformat & share data here.
Anju
Well-Known Member
I wouldn't be thrilled to have a young banker in charge of the country during a global health crisis but looks like number 10 might be regretting Raab taking over. Wonder when they'll send him to sit, PPE free, at Boris's bedside for the day.
Rishi Sunak will take over PM duties if Dominic Raab incapacitated, says No 10
Just worked out that it's only 9 days until my turn
Rishi Sunak will take over PM duties if Dominic Raab incapacitated, says No 10
Just worked out that it's only 9 days until my turn
Last edited:
2hats
Dust.
Moving average from Day-7 to Day-3 (say) for deaths and/or hospital admissions to get an idea of the trend. Everything else (in the public domain anyway) is probably fairly pointless.
Chilli.s
mingling jets and statuettes
Well given that my partner is a nurse, yes I'll pass it on.
They (mostly) are wearing proper masks. The public (mostly) aren't.
SpookyFrank
A cheap source of teeth for aquarium gravel
854
To be expected after the low numbers from the last two days sadly. The general trend is still towards a levelling off of the death rate though. In another week or so we should be seeing a consistent downward trend.
elbows
Well-Known Member
Although I'm not planning to do anything fancy myself, when looking at the data again I suppose the most obvious thing I am struck by is that it might be possible to get a fair estimate of how many days behind the hospital death reality the daily figures actually are. This wont be a constant rate over time but there are certainly some clues about this in the data I share.
elbows
Well-Known Member
Not really. Those numbers are misleading. Thats what my recent tables of data have been about.
Thats not to say that what you are suggesting we will see cannot possibly happen, only that the data has extra lag and unreliability, especially in terms of data for the last 5 days at any moment in time, which will be missing more cases due to reporting delays.
I will try to do a graph showing number of hospital deaths that people are given every day, and what the actual number of deaths on each day in hospital turned out to be once the data improved. I can only do England though rather than the whole of the UK, because I dont have the data to do the others.
(Before I say this, I do appreciate that those who are sick or infirm are not walking about).
Virus aside, my god it’s lovely out there. The sun is shining, it’s like an English summers day, the flowers are out, the trees are budding. People are walking about smiling, talking at a distance. Neighbours are offering to help each other, complete strangers are helpful (when I broke down yesterday on way home after my night shift, a workman went into Wilkos and bought me oil and his pal put it in my car). There’s hardly any cars about, people are cycling. Everyone looks relaxed and so bloody well. The world outside, if you didn’t know there was a pandemic, is about 150% nicer!
Virus aside, my god it’s lovely out there. The sun is shining, it’s like an English summers day, the flowers are out, the trees are budding. People are walking about smiling, talking at a distance. Neighbours are offering to help each other, complete strangers are helpful (when I broke down yesterday on way home after my night shift, a workman went into Wilkos and bought me oil and his pal put it in my car). There’s hardly any cars about, people are cycling. Everyone looks relaxed and so bloody well. The world outside, if you didn’t know there was a pandemic, is about 150% nicer!
elbows
Well-Known Member
OK so the first graph is the number of deaths reported to the public on each day.
The 2nd graph has the deaths actually listed by the day they died, not the day the death was reported.
England only, due to data for other nations not being available to me in the same way. Daily figures, rather than the cumulative stuff I normally go for.
The 2nd graph can improve a lot over time, the first cannot. In particular the 2nd graph reveals how much less refined the data for recent days is, and todays data for actual hospital deaths yesterday is pretty much useless right now. I dont mean that the data quality suddenly reduced in recent days, I mean this is an ongoing phenomenon, the picture of recent days will continue to be very incomplete until the further passage of time.
edited to add - so the 2nd graph is more accurate, but there is lag. It gives a much more correct picture of march, but I cannot really say the same for april yet.
If I didnt make any mistakes with the data then the area taken up by the bars should actually be the same for both graphs, just more accurately distributed in the 2nd one. And the 2nd one makes it much more obvious how incomplete the total for yesterday and several days prior currently is.
The 2nd graph has the deaths actually listed by the day they died, not the day the death was reported.
England only, due to data for other nations not being available to me in the same way. Daily figures, rather than the cumulative stuff I normally go for.
The 2nd graph can improve a lot over time, the first cannot. In particular the 2nd graph reveals how much less refined the data for recent days is, and todays data for actual hospital deaths yesterday is pretty much useless right now. I dont mean that the data quality suddenly reduced in recent days, I mean this is an ongoing phenomenon, the picture of recent days will continue to be very incomplete until the further passage of time.
edited to add - so the 2nd graph is more accurate, but there is lag. It gives a much more correct picture of march, but I cannot really say the same for april yet.
If I didnt make any mistakes with the data then the area taken up by the bars should actually be the same for both graphs, just more accurately distributed in the 2nd one. And the 2nd one makes it much more obvious how incomplete the total for yesterday and several days prior currently is.
Last edited:
elbows
Well-Known Member
One last graph.
This one shows where the new number published today, which is the very long final bar in the first graph in my previous post, actually ends up being distributed on the 2nd graph in that post.
Assuming the same sort of reporting delays continue, when you hear a figure for how many people died in hospital yesterday, this is how those particular deaths were actually likely to have been spread out over time.
This one shows where the new number published today, which is the very long final bar in the first graph in my previous post, actually ends up being distributed on the 2nd graph in that post.
Assuming the same sort of reporting delays continue, when you hear a figure for how many people died in hospital yesterday, this is how those particular deaths were actually likely to have been spread out over time.
Last edited:
elbows
Well-Known Member
I think I picked the wrong day to miss the press conference. WIll have to go back and watch it.
23m ago 17:17
23m ago 17:17
Similar threads
