Any data/math whizzes here?

[tholdren]

Quick do a scatterplot using avg age and death rate.

Then do a scatter plot with avg age and covid death rate.

Let me know what you find out.

Use the same countries.

[midnightpulp]

Bwahahahahahhahahahahahajahah

The only thing you have correct is scattetplots visual. Worst math sense I have seen.

Well besides the time you said states don't include probable deaths.

That was also ignorant

Thanks to RandomGuy for leading me to find ready made linear regression and correlation calculators. Let's see what we found!

Since the calculator flipped the trajectory, the correlation will be negative. I'm sure you know negative doesn't mean reduced correlation or no correlation, right? Def:

Negative correlation is a relationship between two variables in which one variable increases as the other decreases

This will show as stringency increases, deaths decrease. This calculator drew the trend line.



But you keep wanting to see the correlation coefficient right? Got one for you:



.745 out of 1. That is a STRONG correlation.

Lockdowns, at least on DAY OF FIRST REPORTED DEATH, WORK.

[tholdren]

Thanks to RandomGuy for leading me to find ready made linear regression and correlation calculators. Let's see what we found!

Since the calculator flipped the trajectory, the correlation will be negative. I'm sure you know negative doesn't mean reduced correlation or no correlation, right? Def:



This will show as stringency increases, deaths decrease. This calculator drew the trend line.



But you keep wanting to see the correlation coefficient right? Got one for you:



.745 out of 1. That is a STRONG correlation.

Lockdowns, at least on DAY OF FIRST REPORTED DEATH, WORK.

BwahahahahhahahahahahHHHHHHhHHHHHHhahHHHHHHHHHhHHH H

Wrong.

[tholdren]

Thanks to RandomGuy for leading me to find ready made linear regression and correlation calculators. Let's see what we found!

Since the calculator flipped the trajectory, the correlation will be negative. I'm sure you know negative doesn't mean reduced correlation or no correlation, right? Def:



This will show as stringency increases, deaths decrease. This calculator drew the trend line.



But you keep wanting to see the correlation coefficient right? Got one for you:



.745 out of 1. That is a STRONG correlation.

Lockdowns, at least on DAY OF FIRST REPORTED DEATH, WORK.

BwahahahahhahahahahahHHHHHHhHHHHHHhahHHHHHHHHHhHHH H

Wrong.

[tholdren]

this is why you are terrible at math bwahahahahahahhahahah

[midnightpulp]

this is why you are terrible at math bwahahahahahahhahahah

I'll take that as your concession. If you were perceptive and unbiased, you'd see my data makes a case for the Swedish strategy in some respects (problem was, Sweden was too late). I'll let you figure out why.

[tholdren]

I'll take that as your concession. If you were perceptive and unbiased, you'd see my data makes a case for the Swedish strategy in some respects (problem was, Sweden was too late). I'll let you figure out why.

LolIoIloll its not a concession because you are Ignorant to how data is collected. The only way that correlation could be considered causation is if these data points were obtained equally.

The arent. Which is the point and has been the point. You are using data collected differently and trying to compare if they were not. Again just like saying case count raw data is important in measuring the disease.

Its not.

You dont understand this. /thread

[ElNono]

^And Will Hunting brought up a key observation in how the trajectory is parabolic. It seems countries with a 25 or greater lockdown index can manage their death toll pretty well, but once you cross that 20 threshold, the death toll accelerates. This would vibe with the exponential nature of the virus. Going into that 20 index which might mean reopening places like bars, theaters, and churches which causes the death toll to take off.

I would argue it's closer to exponential. In data terms, you're plotting lockdown efficacy vis a vis deaths, but because deaths are generally a subset of infections, if infection rate is exponential, so are deaths.

Now, the criticism one would put in such a graph is that not all countries have the same health system, testing capacity nor approached other mitigating factors the same.

[tholdren]

I would argue it's closer to exponential. In data terms, you're plotting lockdown efficacy vis a vis deaths, but because deaths are generally a subset of infections, if infection rate is exponential, so are deaths.

Now, the criticism one would put in such a graph is that not all countries have the same health system, testing capacity nor approached other mitigating factors the same.

He doesn't understand basic math. Which is the problem with the general public and media.

[ElNono]

He doesn't understand basic math. Which is the problem with the general public and media.

Wrong

[midnightpulp]

LolIoIloll its not a concession because you are Ignorant to how data is collected. The only way that correlation could be considered causation is if these data points were obtained equally.

The arent. Which is the point and has been the point. You are using data collected differently and trying to compare if they were not. Again just like saying case count raw data is important in measuring the disease.

Its not.

You dont understand this. /thread

We well never, ever get "clean" data in this regard. It'll probably take a decade to filter it where it can be "equal." And you know that, which is why you're using it as deflection. "No, no. I won't believe anything until the data is spotless." But we have some data and we can try to equalize it to an extent (me only focusing on EU for this plot, similar healthcare systems, population health, pop density, etc) and make inferences.

[tholdren]

We well never, ever get "clean" data in this regard. It'll probably take a decade to filter it where it can be "equal." And you know that, which is why you're using it as deflection. "No, no. I won't believe anything until the data is spotless." But we have some data and we can try to equalize it to an extent (me only focusing on EU for this plot, similar healthcare systems, population health, pop density, etc) and make inferences.

let's just bump this for fun.

[tholdren]

or are we just ignoring the fact that you implied there is no good data based on how the data is being collected....

I'll wait for the white flag

[midnightpulp]

or are we just ignoring the fact that you implied there is no good data based on how the data is being collected....

I'll wait for the white flag

Nope. I said we don't have spotless data, which is your request. The data is good enough to make inferences, though. And the inference here is .75 correlation coefficient. STRONG.

[midnightpulp]

Bump once more for tholdren.

The relationship between two variables is generally considered strong when their r value is larger than 0.7.

RandomGuy, STRONG

[Will Hunting]

I would argue it's closer to exponential. In data terms, you're plotting lockdown efficacy vis a vis deaths, but because deaths are generally a subset of infections, if infection rate is exponential, so are deaths.

Now, the criticism one would put in such a graph is that not all countries have the same health system, testing capacity nor approached other mitigating factors the same.

Exponential would be a better description, I glanced at the shape of the relationship and said parabolic. Imo it’s definitely an exponential relationship though.

[DMC]

I would argue it's closer to exponential. In data terms, you're plotting lockdown efficacy vis a vis deaths, but because deaths are generally a subset of infections, if infection rate is exponential, so are deaths.

Now, the criticism one would put in such a graph is that not all countries have the same health system, testing capacity nor approached other mitigating factors the same.

Nor do they report at the same accuracy.

[Millennial_Messiah]

How come people haven't mentioned me yet? I'm one of the greatest data science/data analytics minds in the USA.

[Millennial_Messiah]

r= 0.7 (or -0.7) is not strong at all, it's mediocre at best. that means only 49% (less than half) of the data is explained by the least squares regression line

[DarrinS]

I'm perfectly suited to live under "lockdown" forever. I suspect this policy negativity affects certain groups, tho. Probably in a profoundly negative way.

[tholdren]

Nope. I said we don't have spotless data, which is your request. The data is good enough to make inferences, though. And the inference here is .75 correlation coefficient. STRONG.

LolololI'll this chart.

[midnightpulp]

r= 0.7 (or -0.7) is not strong at all, it's mediocre at best. that means only 49% (less than half) of the data is explained by the least squares regression line

Hmmm, seems Joel taught you well. Impressed with your brief description. But this is what I'm reading.

–0.70. A strong downhill (negative) linear relationship

https://www.dummies.com/education/ma...coefficient-r/

Values between 0.7 and 1.0 (-0.7 and -1.0) indicate a strong positive (negative) linear relationship via a firm linear rule.

http://www.dmstat1.com/res/TheCorrel...ntDefined.html

Etc

Here's an interesting thing, though. When I calculate the coefficient by grouping the countries into 3 sets (40-100 stringency, 20-40 stringency, 20-0 stringency) and averaging their death tolls and lockdown stringency respectively, I get a r=.95. I feel this filters the noise a bit, i.e. the wide death toll ranges we have in the sub-20 stringency group.

[tholdren]

Hmmm, seems Joel taught you well. Impressed with your brief description. But this is what I'm reading.

–0.70. A strong downhill (negative) linear relationship

https://www.dummies.com/education/ma...coefficient-r/

Values between 0.7 and 1.0 (-0.7 and -1.0) indicate a strong positive (negative) linear relationship via a firm linear rule.

http://www.dmstat1.com/res/TheCorrel...ntDefined.html

Etc

Here's an interesting thing, though. When I calculate the coefficient by grouping the countries into 3 sets (40-100 stringency, 20-40 stringency, 20-0 stringency) and averaging their death tolls and lockdown stringency respectively, I get a r=.95. I feel this filters the noise a bit, i.e. the wide death toll ranges we have in the sub-20 stringency group.

bwahjhahaaaahahusshshssss

This post

[midnightpulp]

I'm perfectly suited to live under "lockdown" forever. I suspect this policy negativity affects certain groups, tho. Probably in a profoundly negative way.

At least for the initial lockdown phase, my data shows that countries can mitigate death toll with as low as a 25 stringency. But again, I'd like to clarify this is only the initial phase when countries might get a sense something is wrong and close a few things like bars, limit huge gatherings, etc. I haven't examined the efficacy of PEAK lockdown stringency yet.

[tholdren]

At least for the initial lockdown phase, my data shows that countries can mitigate death toll with as low as a 25 stringency. But again, I'd like to clarify this is only the initial phase when countries might get a sense something is wrong and close a few things like bars, limit huge gatherings, etc. I haven't examined the efficacy of PEAK lockdown stringency yet.

bwhahahahahahashshshshshsjsjsjsjsjsjsjsjsususushsu sushsusususususys

My data shows


Bwaahayahaayahaauaasussusuussuss

A scatterplot


Lolololol