Any Idea if this is Accurate?

That's not what the scenario said. It didn't say why he went to the doctor, jus that he did. It's very important for analysis to keep track of what is actually going on.

>It didn't say why he went to the doctor

so why did he go get tested?

Attached: wt98649385743.webm (300x600, 1.1M)

The utility here is about extracting accurate information from imperfectly-accurate empirical investigations. Which is all empirical investigations. You can't get around this limit on certainty by ignoring it.

Because it serves the point of the person describing the scenario. If you can't imagine analyzing the situation without an explanation for that, imagine that people are selected for the test by lottery from the general population.

Is this a question on the patient being sick or the reliability of the test?

its accurate and we use it often in game theory and medicine
its also why you see so many different numbers re: corona- the tests are imperfect and it creates a lot of confusion when crafting reliable estimates

that's what i'm getting at. the cunt wasn't a patient till he went to the dr

Attached: ser6859252.webm (480x320, 2.75M)

Let me clarify: if the tests detects a bacteria in the patient's bloodstream and that bacteria in the quantity of x/mlL is shown to cause sickness then the patient is tested with another tool that relies on another set of probabilistic outcomes and this is why I think that the problem is pretty sloppily presented.

The problem is sloppily presented.
If a doctor knows that 99% of people are sick and 1% of people are healthy, what are the chances that the person that he is talking to is sick...then the answer is 50%.
If we bring words like "tests" and "patient" into the problem then the problem becomes illogical.

Another words: it's the wording of the problem that is a problem not the theorem itself.

P(sick|pos test)=
P(pos test|sick)P(sick)/P(pos test)=
P(pos test|sick)/(P(pos test|sick)P(sick)+P(pos test|healthy)P(healthy))
=0.99*0.01/(0.99*0.01+0.99*0.01)=1/2
seems correct

so if he didn't go to the dr he wouldn't even be sick at all?

Attached: 1540807834029.gif (500x407, 975.8K)

How?
1% of all sick is false
if it's POSITIVE how is it not 99% possible that he's sick?
Negatives don't even matter anymore.

You could go by the logic that probability is meaningless and all things have a constant 50% chance if happening, even a fucking unicorn showing up in Brooklyn.

No. The chances of him being sick would be still 50%.
Now if we are talking about a test, then the test is testing people and some get false positives and some false negatives, right? That's called a reliability of the test.
So we are adding another variable to the problem.
What I'm saying is that the wording of this is sloppy.
Are we asking about the probability of the patient being sick or the reliability of the test aka probability that the test is right/wrong.

Yes, conditional probability is sometimes counterintuitive. What is your point?

>The chances of him being sick would be still 50%.

no. if he didn't go to the dr he would be 100% healthy. like schrodingers cat virus.

Attached: car624968029024.gif (480x480, 2.63M)

>the problem is pretty sloppily presented.
No, you just find it counterintuitive.

>If a doctor knows that 99% of people are sick and 1% of people are healthy, what are the chances that the person that he is talking to is sick...then the answer is 50%.
No. As you reworded it, the answer is 99.99%. Only .01% of those tests would return a false positive.

So what if you test a guy on a deserted island and get a positive, where there was only a pre-existing 1% chance the virus had spread to him from air currents or something?

>what if you test a guy on a deserted island
if you can test him the island isn't deserted and chances are it's the dr who gave him the virus

Attached: em6474.gif (480x265, 2.85M)

No.
Let me try again to explain my point here and my thinking:
If a test tests with 99% accuracy then your chance as an individual to test inaccurately is still 50%.
If 99% of people are sick and you go to a doctor. Your chance is still that you are 50% healty/sick.
Not both.
It's obviously a theoretical question and I'm familiar with this theorem, but the question is sloppy nonetheless.

Not true. Your chance as an individual would be 50% still.

If he's sick it has a 99% of showing positive, but if he's tested positive he doesn't have a 99% chance of being sick since there's so many more healthy people getting tested there will be many false positives.

If you flip a coin 99 times and it lands on heads every single time what are the odds it's gonna be heads on the 100th flip?

i think i might get it now.

i get a letter and a gun in the mail. letter says point gun at my head and pull the trigger. i heard that 99 other cunts got the letter and the gun and survived. but when i go through with it and pull the trigger, the other 99 cunts dont mean nothing cause i only have a 50/50 shot because there are only 2 outcomes for me; click/alive or bang/dead.

can u dig it?

Attached: 1583578738115.jpg (300x287, 41.38K)

See

yeah still be 50/50. i've heard of this one from the 26 black roulette monte carlo thing.

Attached: 1572887703531.webm (540x760, 1.3M)

Just watch this
youtube.com/watch?v=HZGCoVF3YvM
There's a good visual explanation in the vid

>If 99% of people are sick and you go to a doctor. Your chance is still that you are 50% healty/sick.
>It's obviously a theoretical question and I'm familiar with this theorem, but the question is sloppy nonetheless.
You're clearly not familiar with this theorem if you believe this.
If 99% of population are sick and testing accuracy remains the same as in OP then testing positive gives you a 99.9898% you're sick

Not really sure if it applies to OP but we are thinking about it the same way, I think

Dumb gypsy nigger.

Imagine 0% of people are sick. If your test comes up positive, are you still saying it is a 99% chance they're sick? No, it would be a guaranteed false test.

Learn bayes thereom. You sound like a moron. It's not "nerd math," it's actually intuitive.

This is what I see every time someone brings up yellow fever
This exact webm

This is the same as op's post really, only better!

Attached: 1582392942096.jpg (655x527, 36.32K)

this thread sucks

Attached: you should be able to solve this goy.jpg (1125x799, 197.13K)

The tester is a robot thatjust went through the sun where he was sterilized.

also checked

why did you add your assets to your debts?

fuck, irl that took way too long to solve

99 times in a row from a fair coin is something that is vanishingly unlikely to occur over the entire history of the universe. It's far more likely that the coin is unfair than you arrived at that result by chance with a fair coin. Such is the degree of required unfairness for this result to be plausible that a result other than heads is very unlikely for the 100th flip. Realistically speaking, the coin probably has two heads.

Yes, it's correct. You'd have to perform multiple tests to each person if you were to randomly test people in order to reduce the number of false positives.

Hi Yudkowsky, how goes your ongoing struggle to overcome your bias to stuff your fat face with junk food and then try to implement communism?

Attached: dabbing-star-of-david-jewish-funny-hanukkah-holiday-prints.jpg (1500x1500, 649.47K)

Attached: 1464704879-20160531.png (684x1976, 622.25K)

Attached: 1494427145-20170510.png (684x1269, 482.89K)

If there were 0% sick people in the population, a positive result would have 100% chances of being wrong. This probability decreases linearly with the % of sick people in a population, and at 1% sick people you get at 50% chances of being wrong.

>where is the missing $1
A jew stole it

>Yas Forums cannot into Bayes
why do I hang out with you retards again?

I did not verify the calculations in the picture but that's a classic problem in a 101 probability course. It's purpose is to show how counter-intuitive probabilities can be.

ok wait a sec i thought i had it now say i buy a lotto ticket. the odds of winning the lotto is like billions to one but it's actually 50/50 cause i will either win or i won't?

Attached: 1573725984261.webm (480x480, 1.49M)

>i think i might get it now.
You've given no indication of that.

>when i go through with it and pull the trigger, the other 99 cunts dont mean nothing
They don't determine what will happen when you pull the trigger because they are independent events. Whether or not the gun is chambered with the safety off (or whatever) is determined by other factors. It does, however, let you predict what will happen based on the past events in similar situations. This represents knowledge and uncertainty that exists in your head about the state of the situation, not physical indeterminacy of the gun's situation itself, which is set previously.

>i only have a 50/50 shot because there are only 2 outcomes for me; click/alive or bang/dead
Completely wrong in a way that comes out of left field and has no relation to the setup as presented. For one, there are a lot more than two options. Maybe your aim is terrible, and you don't fatally injure yourself. The vital parts of the head are smaller than most people think, and a lot of headshots are non-fatal. The pop-culture gun-to-head suicide posture aims at a spot that's likely to horribly maim you rather than outright kill you, for example. Another of the countless possibilities is that the Russian Roulette aspect is a ruse, and the gun itself is coated with a nerve agent that kills you a few minutes after you handle it. You generally can't infer that if there are X possibilities, each one has a 1/X chance of happening, even if there really are X possibilities. For example, in classic Russian Roulette, one chamber in six is loaded, resulting in a 5/6 chance of survival after the spin, despite there being "two" options.

there are two possibilities. sick or not sick so 50% chance of either being the case.. the size of either group is irrelevant

No

sauce

This oversimplifies the problem. It takes the percentage of people that actually are sick and the percentage of people given false positives and essentially says that, given random sampling, a positive is likely to actually be a false positive given the low amount of people who actually are sick.

It falls apart because it assumes the people being tested are random samples and represent the broader population. In reality, someone getting tested for an illness probably already suspects they have it. That means using the probably of being sick among broader population is invalid, because the correct population to use would be the population of people who think they are sick and thus get tested. Ergo the correct proportion to use is the population of people who get tested who actually are sick. The fun thing is that that ratio is already built into the test, which is to say that the 99% probability is correct.

>I don't feel it being right.

t. Every student in their first stats course, ever

Wait till I show you the Monty hall problem! Brianlet

>I reject your problem and substitute my own.
>That means I'm right.