Coronavirus - General Chat - No statistics

[servodude]

It might just be noise and that some samples "appeared early" as the figure from 3 days ago (764) was 11% more than I would have expected (yesterdays 792 was about 6% above)

If you are up for some work then looking at the regional curves would be interesting. Although any exponential data would dominate given sufficient time, what is actually happening in each NHS region is particularly relevant to the question of what merit there is in England wide restrictions.

Sounds like a plan - I mumbled on the modelling thread that I was tinkering with python
Should be easy enough to parameterise what I have and iterate the fetches if I can find out what their API expects for the area filter

To that end:
- do you know where I can find a list of the areaNames as used by the API? (all the docs I've seen so far use England exclusively as an example - and it's wine-o'clock here)

- sd

[johnhemming]

You can get them from here:
https://coronavirus.data.gov.uk/healthc ... %20England

Just go through the drop down box.

[XFool]

Article in the Guardian about the "Barrington Declaration" https://www.theguardian.com/commentisfr ... i-lockdown
Contains:

The science is clear: attaining herd immunity to coronavirus via uncontrolled infection is a fringe view, peddled by a minority with no evidence to back up their position.

Scott.

http://fringereview.co.uk/review/edinbu ... galileo-2/

This outstanding one man show is set around the trials of Galileo in 1633 in which he was tried for heresy after demonstrating that the world circled the sun rather than vice versa.

That was science vs a religious ideological worldview. Here it's science vs a political ideology. Choose your side...

The use of the word "fringe view" to criticise a scientific argument should not apply in science.

Nonsense. Quackery is still quackery and politics is still politics.

Human nature has not changed that much over time.

Absolutely.

[XFool]

Actually, after looking at it, this reference is more interesting than one might expect!

http://fringereview.co.uk/review/edinburgh-fringe/2015/the-trials-of-galileo-2/

This outstanding one man show is set around the trials of Galileo in 1633 in which he was tried for heresy after demonstrating that the world circled the sun rather than vice versa.

"Galileo probably was a genius and a painstaking and careful scientist. However, in his battle with the church he failed to understand that it wasn’t a question of proof, but of politics. Something the Vatican cardinals were expert at."

As an aside I'd like to say I'm not sure Galileo did fully demonstrate "that the world circled the sun rather than vice versa" - not that for a moment I doubt he was correct! He strongly argued for his belief in it, but surely it wasn't convincingly demonstrated until Kepler's analysis and insight based on Tycho Brahe's data (who himself wasn't convinced - for seemingly sound reasons at the time)?

https://en.wikipedia.org/wiki/Galileo_Galilei#Controversy_over_heliocentrism

[XFool]

Belgium facing 'tsunami' of Covid infections, health minister says

https://www.theguardian.com/world/2020/oct/19/belgium-facing-tsunami-of-covid-infections-health-minister-says

Frank Vandenbroucke says situation in parts of country is ‘most dangerous in all of Europe’

[johnhemming]

The use of the word "fringe view" to criticise a scientific argument should not apply in science.

Nonsense. Quackery is still quackery and politics is still politics.

The point about a view is its scientifically validity does not depend upon the number of people who support it.

Hence the word "fringe" is not a valid criticism from a scientific perspective.

[XFool]

The use of the word "fringe view" to criticise a scientific argument should not apply in science.

Nonsense. Quackery is still quackery and politics is still politics.

The point about a view is its scientifically validity does not depend upon the number of people who support it.

Hence the word "fringe" is not a valid criticism from a scientific perspective.

The point about a fringe view is why so few support it, plus where is the evidence (likely related to why so few support it), who supports it and why?
These are all valid questions.

If a "fringe" view is valid, the weight of evidence, in time, will lead to its adoption so it will no longer be a "fringe" view. Unfortunately, in the midst of a pandemic, we don't have the luxury of time.

But there are ALWAYS (quack) "fringe" views. They hardly ever go away. Andrew Wakefield is as popular as ever on the net. As you said yourself "human nature doesn't change".

One simply has to apply some judgement. With say The Barrington Declaration, the obvious was obvious from day one. "Science" doesn't generally get done or get verified by osteopaths and people calling themselves professor Buckethead or whatever. And yes, I know "anybody could sign" etc. That's a clue from the get go. IMO.

If I were a follower of pyad (!) I might talk about the "smell test".

[dealtn]

That was science vs a religious ideological worldview. Here it's science vs a political ideology. Choose your side...

Actually some of it is Science vs Science, and there is even some Political Ideology vs Science, as well as the Science vs Political Ideology you (exclusively) claim. However by your own admission you aren't prepared to read stuff sufficiently far from your own confirmation bias to have picked up on that.

[johnhemming]

The point about a fringe view is why so few support it

The number of people who support something should not be part of whether or not it has scientific validity.

[XFool]

The point about a fringe view is why so few support it

The number of people who support something should not be part of whether or not it has scientific validity.

But what I said was:

The point about a fringe view is why so few support it, plus where is the evidence (likely related to why so few support it), who supports it and why?

[johnhemming]

The point about a fringe view is why so few support it

The number of people who support something should not be part of whether or not it has scientific validity.

But what I said was:

The point about a fringe view is why so few support it, plus where is the evidence (likely related to why so few support it), who supports it and why?

I think we just have to disagree here. I think it is worth considering various options even the unpopular ones and looking at the evidence. That something is fringe is not relevant to the question as to whether it is scientifically valid.

[XFool]

I think we just have to disagree here.

Agreed!

I think it is worth considering various options even the unpopular ones and looking at the evidence.

It likely is, and somebody, somewhere, sometime likely will do just that. They may well be doing it right now. But, until and unless that results in a clear and definite conclusion, it will have little practical value.

For myself, I am certainly not in any position to decide between the current consensus and an alternative one. So I don't try to.

That something is fringe is not relevant to the question as to whether it is scientifically valid.

In the more general sense that you likely mean that, I agree. However, there are good reasons why some views are "fringe" and should remain so.

[johnhemming]

For myself, I am certainly not in any position to decide between the current consensus and an alternative one. So I don't try to.

There are, of course, quite a range of views about the virus even amonst people who understand the science.

I find it useful to have a good idea as to what is going to happen. Then one can make judgments as to how to respond to that.

[servodude]

You can get them from here:
https://coronavirus.data.gov.uk/healthc ... %20England

Just go through the drop down box.

Thanks I scraped them from the request to load the dropdown

Here's what the last 44 days look like across the regions given
- for each there are two best fit exponential trends (ae^(bx) over 30 days data
- the green is that for the data ending at 14 days ago
- the purple (dashed) for that ending at 7 days ago
- the blue bit on each graph is the data since the end of that used to calculate the later trend (basically showing the last week vs the two previous "predictions")

the intent is that the difference between these "trends" might show if things are changing under the covers
- but with very low numbers in play this might get drowned in random events



If you fancy a go yourself - here's what I used

Code: Select all

# -*- coding: utf-8 -*-
from requests import get
from json import dumps,loads
from urllib.parse import urlencode
from scipy.optimize import curve_fit
import numpy as np
import matplotlib.pyplot as plt

ENDPOINT = "https://api.coronavirus.data.gov.uk/v1/data"
AREA_TYPE = "nhsRegion;"

# function def for curve fit
def exponential(x, a, b):
    return a*np.exp(b*x)

fetches = [
    {'area':'East of England'          ,'row':0,'col':0    },
    {'area':'London'                   ,'row':0,'col':1    },
    {'area':'Midlands'                 ,'row':1,'col':0    },
    {'area':'North East and Yorkshire' ,'row':1,'col':1    },
    {'area':'North West'               ,'row':2,'col':0    },
    {'area':'South East'               ,'row':2,'col':1    },
    {'area':'South West'               ,'row':3,'col':0    }
]

# fudge the display to prevent overlapping
fig, ax = plt.subplots(4,2, dpi=1200)
fig.tight_layout(rect=[0, 0.03, 1, 0.95])
fig.suptitle('Hospital admissions')

for fetch in fetches:
    filters = [
        f"areaType={ AREA_TYPE }"
        f"areaName={ fetch['area'] }",
    ]
   
    structure = {
        "name":"areaName",
        "date": "date",
        "admissions":"newAdmissions",
    }
   
    api_params = {
        "filters": str.join(";", filters),
        "structure": dumps(structure, separators=(",", ":"))
     }
   
    api_params["format"] = 'json'
    response = get(ENDPOINT, params=urlencode(api_params), timeout=10)
    assert response.status_code == 200, f"Failed request for: {response.text}"
    ad = response.content.decode()
   
    # convert out from JSON
    data = loads(ad)['data']
    # extract and reorder latest samples
    points = 44
    expwin = 30
   
    latest = data[:points]
    latest.reverse()
   
    # extract values from tuples
    actual = [x['admissions'] for x in latest]
   
    # create ranges for calcs and plot
    x_full = list(range(0,points))
    x_extend = list(range(10,points+7))
   
    # plot original data
    ax[fetch['row']][fetch['col']].plot(x_full, actual)
       
    # define styles
    offsets = [
        {'val':14, 'st':''},
        {'val':7, 'st':'--'}
        ]
 
    for off in offsets:
        # extract set
        stop_p = len(latest) - off['val']
        start_p = stop_p - expwin
        part = [x['admissions'] for x in latest[start_p:stop_p]]
        x_ax = list(range(start_p,stop_p))
        # curve fit
        re = curve_fit(f=exponential, xdata=x_ax, ydata=part)
        #extrapolate
        a=re[0][0]
        b=re[0][1]
        y_plt =  [exponential(x,a,b) for x in x_extend]
       
        # overlay on data plot
        ax[fetch['row']][fetch['col']].plot(x_ax, part )
       
        ax[fetch['row']][fetch['col']].plot(x_extend, y_plt, off['st'])
        ax[fetch['row']][fetch['col']].set_title(fetch['area'])

ax[3,1].axis('off')

plt.show()
fig.savefig('eng_admissions.png')


forgive the hacky-ness of it - I'll blame the shiraz

- sd

[johnhemming]

Thanks I scraped them from the request to load the dropdown

Thanks. Do you easily have a doubling time for your best fits?

[johnhemming]

Hospital admissions back up again to 785 from 632 previously 792. Big jump in NE from 159-210 and NW from 196-272.

[XFool]

Understanding how ‘overdispersion’ works is key to controlling Covid

The Guardian

As few as 10% of people are responsible for 80% of transmission – and that must shape how we tackle this virus

[johnhemming]

Understanding how ‘overdispersion’ works is key to controlling Covid

The Guardian

As few as 10% of people are responsible for 80% of transmission – and that must shape how we tackle this virus

This is sadly something that should have been more effectively studied at the start.

[langley59]

'I think it’s unlikely that we will end up with a truly sterilising vaccine – i.e. something that completely stops infection – and it’s likely that the disease will circulate and be endemic,' Sir Patrick said in a meeting of the Lords' National Security Strategy Committee this afternoon.

Finally...so can we treat it like the flu and now get on with life again please?

[Mike4]

'I think it’s unlikely that we will end up with a truly sterilising vaccine – i.e. something that completely stops infection – and it’s likely that the disease will circulate and be endemic,' Sir Patrick said in a meeting of the Lords' National Security Strategy Committee this afternoon.

Finally...so can we treat it like the flu and now get on with life again please?

Whenever we manage work out how to stop it filling our ICUs to overflowing, yes.