South African Covid-19 News and Discussions 2

Status
Not open for further replies.
tetrasect

tetrasect

Executive Member
Joined
Aug 22, 2009
Messages
9,105
Gordon_R said:
Calculating a proper number for Rt is much, much harder than it looks. We know that testing in SA is neither random, nor consistent, and misses about 90% of all infections. Proper estimation in such a scenario involves looking both at new test cases, as well as deaths, and modelling the underlying number of infections based on various assumptions.



Thanks for investigating. Noise in data takes many forms, and what you found is all part of what I was referring to.
Click to expand...

Pretty sure they are just running the official numbers. Guesstimating actual infections does not make sense in this context. Whether 10% or 100% of cases are detected, the Rt is the same.
 
Gordon_R

Gordon_R

Honorary Master
Joined
Jul 5, 2009
Messages
20,817
tetrasect said:
Pretty sure they are just running the official numbers. Guesstimating actual infections does not make sense in this context. Whether 10% or 100% of cases are detected, the Rt is the same.
Click to expand...

I am sure they are only running the official numbers, but this does not mean it is the 'right' way to measure Rt. If you have whole provinces like the W-Cape not testing under-55s, that can skew the data considerably.

This international site gives a broad perspective on methodological issues:
What is the difference between "positives" and "test-adjusted positives"? Why don't the lines match up when I "show cases"?
Our model attempts to correct for testing volume. The dotted black line shows actual new cases reported, the blue line shows what the model believes actual cases would have been if you correct for testing volume. If today you tested 100 people and 10 came back positive and tomorrow you tested 500 people, you might assume you'd get 5x the number of positives back (all things being equal).

For instance, California shows a marked increase in cases over most of June, but the blue test-adjusted curve doesn't react nearly as much. This is because testing volume has ramped significantly in California in the same time period. If there are more tests, there will be more positives. The model works off the blue test-adjusted curve rather than the black dotted curve. Doing so ensures that we're looking at the 'true' trend in cases. That being said, tests are not a random sample of the population and therefore it's possible (and likely) that the selection of this group changes over time to include many more people who are not symptomatic and who are simply getting a test as a precaution. This means that if anything, the blue curve may be understating the true infections going forward. Please keep this in mind when drawing conclusions.

How does the new model work?
In the simplest terms, it searches for the most likely curve of Rt that produced the new cases per day that we observe. It does this through some neat (and powerful!) math that is beyond the scope of this FAQ. In more complex terms: we assume a seed number of people and a curve of Rt over the history of the pandemic, we then distribute those cases into the future using a known delay distribution between infection and positive report. We then scale and add noise based on known testing volumes via a negative binomial with an exposure parameter for a given day to recover an observed series. We plan on publishing our code soon, so if you’re so inclined you’ll be able to run it, too.

Why do you use positive tests instead of hospitalizations or deaths to inform your model? Aren't the latter two far more reliable?
In general, hospitalizations and deaths are more reliable than tests to see the true Rt curve. However, they are also both time-shifted fairly dramatically from the time of infection. As of this time we have not included them in our model, but we are considering ways to reliably and accurately include them to ensure the model is as accurate as possible.
Click to expand...
 
Geoff.D

Geoff.D

Honorary Master
Joined
Aug 4, 2005
Messages
26,878
Gordon_R said:
I am sure they are only running the official numbers, but this does not mean it is the 'right' way to measure Rt. If you have whole provinces like the W-Cape not testing under-55s, that can skew the data considerably.

This international site gives a broad perspective on methodological issues:
Click to expand...
The work being done by the team uses the same basic model as that used by the USA group.

They are constrained by this statement, from description of the technique they are using:

In South Africa we are unable to obtain such data because the government won't release it. So to proceed we rely on work done using USA data for the rt.live dashboard where an empirical distribution of onset delays was obtained from US case data.
Click to expand...
 
Paulsie

Paulsie

Executive Member
Joined
Apr 6, 2020
Messages
5,486
Chris_the_Brit said:
Some newly updated regulations from the tourism briefing:

> Curfew moved to 10pm
> Intra-provincial (travel within a province) is allowed for leisure purposes.
Click to expand...
A friend of mines father is a treasurer with the local municipality. According to him the government has run out of money to be able to assist many. As such they are apparently planning to open everything up on 15th August. That's why TERS uif was extended until then (which was originally viewed as an odd date). Fingers crossed.
 
squirrel

squirrel

Expert Member
Joined
Jun 11, 2005
Messages
1,979
Can at least explore the Western Cape a bit now. Bit of breathing after being in an apartment for months. I'm definitely checking out one of these glamping spots!
 
  • Like
Reactions: air
Temujin

Temujin

Honorary Master
Joined
Apr 18, 2015
Messages
18,313
Paulsie said:
A friend of mines father is a treasurer with the local municipality. According to him the government has run out of money to be able to assist many. As such they are apparently planning to open everything up on 15th August. That's why TERS uif was extended until then (which was originally viewed as an odd date). Fingers crossed.
Click to expand...
NDZ and Co will not give up power
 
ToxicBunny

ToxicBunny

Oi! Leave me out of this...
Joined
Apr 8, 2006
Messages
113,505
Paulsie said:
A friend of mines father is a treasurer with the local municipality. According to him the government has run out of money to be able to assist many. As such they are apparently planning to open everything up on 15th August. That's why TERS uif was extended until then (which was originally viewed as an odd date). Fingers crossed.
Click to expand...
Can we tar and feather the friend and his father when they are wrong?
 
Arzy

Arzy

Honorary Master
Joined
Apr 18, 2004
Messages
29,019
Paulsie said:
A friend of mines father is a treasurer with the local municipality. According to him the government has run out of money to be able to assist many. As such they are apparently planning to open everything up on 15th August. That's why TERS uif was extended until then (which was originally viewed as an odd date). Fingers crossed.
Click to expand...
Don't recall which thread I posted it in my my response to the IMF deal being done was that now that the government has secured all these loans they are just going to drop all the lockdown ****.

Not expecting it to be on for much longer.
 
ToxicBunny

ToxicBunny

Oi! Leave me out of this...
Joined
Apr 8, 2006
Messages
113,505
Azg

Azg

Expert Member
Joined
Jul 15, 2013
Messages
3,213
Arzy said:
Don't recall which thread I posted it in my my response to the IMF deal being done was that now that the government has secured all these loans they are just going to drop all the lockdown ****.

Not expecting it to be on for much longer.
Click to expand...
Surely losses in tax revenue and the economy outweigh the loans? In which case it does not make sense to say the govt lockdown strategy was intended to get loans.
 
Arzy

Arzy

Honorary Master
Joined
Apr 18, 2004
Messages
29,019
Azg said:
Surely losses in tax revenue and the economy outweigh the loans? In which case it does not make sense to say the govt lockdown strategy was intended to get loans.
Click to expand...
So that they have bulk (R100b) suddenly added to the purse to issue out in tenders and aid isn't a strategy?
 
tetrasect

tetrasect

Executive Member
Joined
Aug 22, 2009
Messages
9,105
Chris_the_Brit said:
https://twitter.com/i/web/status/1288746301463683072

Will other provinces be so transparent? :unsure:
Click to expand...


Will leave the fine tooth combing to others, but just glancing at that list; some of those purchases look very overpriced.

"Masiqhame Trading 1057" charged between R72.63 and R84.74 per pair of household gloves?
4643 pairs(hopefully) of gloves for R411,996.-???

Just searched google for "household latex gloves" and the first link that came up was this https://www.cashbuildonline.co.za/products/latex-protective-household-gloves

R21.95 per pair and that's retail. Wholesale pricing should be WAY cheaper, especially when they are importing the stuff from China.

So even just by purchasing from the first local link on google they could have saved R310,082.- of the R411,996.- spent.

Go on Alibaba.com and in the first row of results for "latex household gloves" I found 5000 pairs for $170.-

That's R14,265.- (excluding shipping which would be a couple thousand extra).

And we spent R411,996.- ???????????????????????

WTF is this ****???!!!!!!!! o_O:mad:
 
Last edited:
Status
Not open for further replies.
Top