@terrycream
How long do you expect that tp be valid?
The us is 1.61*1.27^t according to ponderable
@eladar saidThis depends on the suppression strategy and whether social distancing can be effectively implemented. Suppose the measures the government is attempting to take in the UK are successful, then we're on the green curve and it should start going out of the exponential phase at some point in early-mid April, so in about 21 days. However, in the event of total policy failure we're on the black curve and it's looking exponential until early May so for 40 days from today. This is based on visual inspection of their graphs, so there's a huge degree of error here. The graph for the US in the appendix is looking similar.
@Ponderable
What is your some time? Do you think it will be accurate until day 60?
https://www.imperial.ac.uk/media/imperial-college/medicine/sph/ide/gida-fellowships/Imperial-College-COVID19-NPI-modelling-16-03-2020.pdf
@eladar saidI think, but I'm not sure, that a test for the presence of anti-bodies is not available yet. They are working on it. As an exercise in epidemiology and seeing what the likely prevalence of the disease is it would be useful, but this comes with a caveat. The tests aren't perfect, they're something like 76% accurate [1]. As confirmation of a diagnosis this isn't a problem, as the medics understand the toolkit, and at an epidemiological level it's not a problem as they can correct for this in their figures. But at an individual level it can cause problems. The articles don't say what the direction of error is, but if a negative is a definite negative and a positive is a possibly positive then that's fine - a worry for the patient, but the testing staff can explain that to them. The other way round and it can be a catastrophe. If the test produces definite positives but only possible negatives then the person being tested can be told they've tested negative but actually have the disease and go back to work and infect a bunch of people.
If you wanted to do it, you could take a random sample of people and see if they are sick or have the antibodies. Then you can get an estimation of percent who have the virus.
But they have other things to worry abput at the moment.
[1] https://en.wikipedia.org/wiki/COVID-19_testing#Chest_CT_scans
@DeepThought
This depends on the suppression strategy and whether social distancing can be effectively implemented. Suppose the measures the government is attempting to take in the UK are successful, then we're on the green curve and it should start going out of the exponential phase at some point in early-mid April, so in about 21 days
Seeing as this is day 24 you are saying it will likely be valid until day 45 at the very least.
So you think we will have about 80 thousand dead by the time we stop seeing exponential growth.
@eladar saidThis is based on visual inspection of the graph, so plus or minus 20,000, but yes that kind of figure.
@DeepThought
This depends on the suppression strategy and whether social distancing can be effectively implemented. Suppose the measures the government is attempting to take in the UK are successful, then we're on the green curve and it should start going out of the exponential phase at some point in early-mid April, so in about 21 days
Seeing as this is day 24 yo ...[text shortened]... .
So you think we will have about 80 thousand dead by the time we stop seeing exponential growth.
@deepthought saidSo your confidence interval for estimating total US deaths by mid April is 60 to 100 thousand.
This is based on visual inspection of the graph, so plus or minus 20,000, but yes that kind of figure.
@deepthought saidCorrection to the above - the UK is reported as having ordered 3.5 million tests which look for the anti-body, so that type of testing is available.
I think, but I'm not sure, that a test for the presence of anti-bodies is not available yet. They are working on it. As an exercise in epidemiology and seeing what the likely prevalence of the disease is it would be useful, but this comes with a caveat. The tests aren't perfect, they're something like 76% accurate [1]. As confirmation of a diagnosis this isn't a probl ...[text shortened]... d infect a bunch of people[/i].
[1] https://en.wikipedia.org/wiki/COVID-19_testing#Chest_CT_scans
@DeepThought
So, as I'm adding data points to the logistical regression it crashes. Singular Matrix. Do you have any idea why increasing the constraints causes the coefficient matrix to not invert? Is it perhaps just too bad of an approximation? With least squares on a polynomial fit we can always get a result no matter how bad it fits the data, correct? What is fundamentally different other than the equation form...an equation with some asymptotic behavior?
@eladar saidI think up to about day 42, then the measures should be effective.
@Ponderable
What is your some time? Do you think it will be accurate until day 60?
@ponderable saidToday is day 26, so 16 more days.
I think up to about day 42, then the measures should be effective.
Wow, the day you used when you first gave your equation.
38,416
@joe-shmo saidHey, just going through this thread and realized this post may have sounded like I was saying don't follow Deepthoughts analysis.
Don't bet on the previous model, much safer to go with my latest revision.
For Clarification: at the time I was stating don't follow MY previous model predictions. I was figuring the more data coming in, the better it will be for the regression. That is not how its turning out. The Logistical regression is failing with the current data set. I would put very little stock in it.
@Eladar
I think I figured out the issue. The computation is crashing when it is working with numbers in the high hundreds/thousands ( its probably trying to do exponentials with those numbers and failing ). If I divide all the Y data by 10, it produces a curve. I just have to figure out how the actual curve relates to the adjusted curve. I suspect since I'm not adjusting the independent variable it will just be a matter of scaling the "y values" of the altered curve by 10.
So, the latest curve is ( multipled by 10 which just effects "a" and "d" ):
a = 9,778
b = 19218.53
c = -0.307063
d = 15.6
It predicts an inflection of April 2 with 5000 Deaths. Total Deaths anround April 19th 10,000
I should note that If you look back, my numbers are changing significantly. @Deepthought was obviously correct about very large error bars.
I can't help noticing that more Southern and warmer places have been hit harder in Europe and the more Northerly places less hard. This implies to me that the virus is not harmed by warmer weather and the effective social distancing in Winter is what's held up the virus in the meantime. That might also explain why the outbreak was more easily controlled in China.