Look all you need to know is Bayesian inference, everything else is window dressing
this post was submitted on 16 Jan 2024
59 points (100.0% liked)
askchapo
23053 readers
321 users here now
Ask Hexbear is the place to ask and answer ~~thought-provoking~~ questions.
Rules:
-
Posts must ask a question.
-
If the question asked is serious, answer seriously.
-
Questions where you want to learn more about socialism are allowed, but questions in bad faith are not.
-
Try !feedback@hexbear.net if you're having questions about regarding moderation, site policy, the site itself, development, volunteering or the mod team.
founded 5 years ago
MODERATORS
My priors are that you're full of shit. Checkmate
updating. P(not full of shit) = p(full of shit) - epsilon / some shit
Yep still full o shit
Please tell me this is a shitpost because this is the stuff Yuddites use to justify their weird shit
There were yuddites out there talking about bayesian statistics?
I don't even know how to explain it (Link is catalogue of people talking about and "sneering" at (hence the name) the weird shit Yudd and his cultists get up to)
A Reddit link was detected in your comment. Here are links to the same location on Teddit and Libreddit, which are Reddit frontends that protect your privacy.
Don't worry, nobody understands what the fuck is a degree of freedom.
Or statistics in general
hit them with the "wow, normal distribution..." they will get it and you will be so popular
I get it, always hated statistics. Never quite figured how to reason about samples vs populations. Your best bet is to use all of it in practice. Try to understand why each term has that name. You're not trying to be a stastician so you just need to pass.
Call it a Gaussian distribution to sound fancy.
Stats can move pretty fast that's true. I think the way up stay on top of it is to read before the lecture, use the lecture to take notes and understand what the teacher wants you to focus on, and to do the homeworks (possibly extra homework) to use repetition to force it into your brain.
So basically... more work at the stuff you already knew to do lol. Shit sucks.
On the plus side this stuff is kinda useful to know in that you'll be able to more critically read stats in the lying liberal media.
if its any comfort at all, I passed my intro to statistics subject in first year uni with no mathematical experience past year 8
I know it's stressful, but I'm sure you've got a better understanding than you'll give yourself credit for. Goodluck!
For basic stats you can ignore a bunch of the underlying theory and just memorize a couple of formulas (or how to apply those formulas if you get a notes sheet on the test) and get good at finding numbers in tables.
The normal distribution represents the idea that most sets of data tend to cluster around the mean (if you grab someone at random, you are more likely to find someone who's 5'8" than 6'7").
Degrees of freedom is the number of observations in your sample minus 1 and is used to look up a value in a table.
Degrees of freedom is the number of observations in your sample minus 1 and is used to look up a value in a table.
doesn't this lead to a lot of extraneous variables that are actually linearly dependent on a smaller set? or worse -- overconstraining?
This isn't a universal definition of degrees of freedom, it's just "degrees of freedom as it applies in an undergraduate level stats course," which is typically for the t distribution. It's n-1 because you assume all your observations are independent of one another. In other contexts (ANOVA, e.g.), the calculation is different.
I get that, just thought it was a strange definition given what degrees of freedom actually refers to
It might help to understand degrees of freedom in a more general context. In a physical system, the number of degrees of freedom is the number of things that have to be specified for me to know the complete state of the system--for me to know "everything there is to know" about it.
Imagine I've got a featureless point particle floating in a box. How many numbers do I need to write down to specify everything there is to know about the system? Well to begin with, I need to know where the particle is in the box, so I need three numbers: one for the x dimension, one for y, and one for z. That's three degrees of freedom.
Does that tell me everything there is to know? If the system is an unchanging snapshot sure, but not in a real dynamical system. If the particle can move, we need three more numbers: one for its velocity in x, velocity in y, and velocity in z. If you have those three numbers, you can predict where the particle will be at any given time, assuming you know where it started.
So, we'd say this particular system has six degrees of freedom: position in x, y, and z, and velocity in x, y, and z. There are six parameters that can vary to make a unique state of the system, given the constraints we've put on it.
I did stats and just barely passed by cheating
Normal distribution is just the bell curve right?
= {[
]")
]²}^{
} / {
, 
, n, n, n}
I had a Stats professor from China and she was somewhat hard to understand, and we spent a week learning to calculate "Elo" and I had no idea what it was, just had to wrap my head around these equations without knowing what it even represented.
The next week she was showing a variable that could go down to zero, in which case that variable "didn't matt" and we could simplify the equation.
She kept saying "didn't matt"
I realized that she had trouble saying "er" because that's a famously weird and difficult sound to make, and only shows up in English and a couple other languages... So she didn't even try at all, she just skipped over that sound whenever it came up.
Then it hit me. Elo. We had been learning to calculate FUCKING ERROR
I feel understood, I’m unironically going to have a breakdown the next time I’m taught some shit named like “Gretchen’s integrated crosscube” that I have to file into the thousands of of other microconcepts that will all be on my final