🇳🇱
Everything can manifest as a normal distribution under certain circumstances, its called the Central Limit Theory. It isnt that something is "hiding" as a normal distribution, its more tht you created a situation where the data is legitimately normally distributed under your analysis.
There are many many types of distributions and no educated person on the matter would suggest everything is a normal distribution.
@freemo @funny Tegnell working under Gaussian assumptions: "In the normal distribution of a bell curve asymptomatics sit at the margin, whereas most of the curve is occupied by symptomatics, the ones that we really need to stop."
Hint: pandemics are not Gaussian
https://www.nature.com/articles/d41586-020-01098-x?fbclid=IwAR2ELm8oXSN9c8f51ueBpKchykmu1h2NLqu0JXntNCO-XulDB8jaf_zs1KI
No, we debunked that idea in the last message where I addressed the Central Limit Theorem, why are you regurgitating shit when we already covered rather plainly why it makes no sense to say that?
The Central Limit Theorem proves that**everything** is gausian when measured in a particular way, even if it is non-gaussian when measured in a different way. I showed earlier how dice when sampled one at a time have a uniform distribution, yet when sampled as summed groups become gaussian. Literally everything becomes gaussian when sampled this way.
So from that example it should be clear why a statement as naive as "Dice are not Gaussian" is absolutely false, like everything else they can be gaussian, but they can also have other distributions applied, depends entirely on how you organize the data you get from it which applies.
So similarly we know as an indisputable law, that "pandemics are not Gaussian" is an outright lie, as we know **everything** is gaussian when measured/organized in a way that it presents as such, and is entirely accurate and valid within that context (just as the gaussian distribution of the dice accurately predicted future rolls).
If you want to make a useful case of something, which you seem to want to to do but are dancing around the bush so as to avoid making a direct provable (or disprovable) statement, well you need to be more specific.
It isnt about how large the numbers are. Dice rolled one a time are uniform even if i repeat it 10 million times. Dice rolled in small groups of 2 however immediately become gaussian and can be seen even with small samples.
random twitter post (which dont disagree with anything I just said mind you) without you actually stating what your point is doesnt really help with anything. As i stated before, if you want to make a point, make it.
it feels like maybe you dont understand the topic and are just sharing random links because a few words sound relevant or something.. I dunno if you dont explain your point when you share links they are just noise.
im not reading link pasta, stop replying with links unless you have something to say that the link is relevant to.
Why roll your eyes? understanding the CLT is pretty important in my line of work and not knowing it would cause a lot of harm.
As for not always using a gaussian distribution, I would say, duh. We have hundreds of distibutions for a reason and of course you should use the one that fits the situation. I never suggested anything of the sort. However the gaussian is a **special** distribution that crops up a lot more often because of the CLT, which explains why it dominates most discussions, and that is not overuse but perfectly normal.
In short anyone trained in this stuff will use all sorts of distributions and will use the correct one for the situation (which is usually not hard to determine)... but in doing so Gaussian will inevitably be the most common thanks to the CLT, not sure how anyone could disagree with that.
All that said, 1) it sounded like he was making some weird coronavirus segway which he didnt elaborate on 2) not sure why he is arguing that there are other distributions, that was never up for debate 3) if thats what he wants to say he should, you know, say it, because right now its just linkpasta with no point and whatever he is arguing against looks like is not even what anyone is saying anyway.
maybe, but unless he speaks up and explains specifically how he thinks gaussian was misused wrt coronavirus its just noise.. what data/numbers does he think a gaussian was misapplied to when talking about coronavirus? Hell is that even what he is saying, who knows he just keeps posting links.
@js290
If you want an example of the Central Limit Theorem and more specifically why it isnt really fair to say one distribution "hides" in the normal distribution but rather why simply manipulating the circumstances gives rise to a valid normal distribution we can just look at dice as a simple example.
A single die with 6 sides has a uniform distribution among its outcomes, not a normal distribution. In other words, when I roll every side has an equal chance of being the outcome. So rolling a single fair dice has a uniform distribution of outcomes.
However if we **change** the situation and now instead we roll 100 dice at once and add together their values, doing that many times to determine the outcome, with each individual 100-die roll and its sum being the outcome... all of a sudden those outcomes **would** be normally distributed. This is the central limit theorem at work.
No one would say the dice in the second situation is simply the uniform distribution "hiding" as the normal distribution. The situation is different, the experiment is different, it is, legitimately a normal distribution. Its just that the normal distribution arises from **any** other distribution and is the very reason why it is so common. Either way though the normal distribution will accurate predict the outcome in situations it does arise.
@funny