Talk:Bayesian

From RationalWiki
Jump to: navigation, search

This is poorly-written, does a hit-job on Frequentism, and conflates the various senses of "Bayesian". Bayes' Theorem is an utterly banal statement of conditional probability, Bayesian statistics has a number of schools of thought ranging from completely pragmatic (ie it's just one set of tools among many) to completely doctrinaire (fully subjective Bayes as done here - this is uncommon nowadays), and Bayes-as-rationality is quite controversial and something quite different from the rest (well, not so different from fully subjective Bayes). It seems partly cribbed from LessWrong, which is not a very good source. Scrapping with a complete rewrite would be in order, but that's too much work for me.173.17.100.187 (talk) 02:16, 30 August 2014 (UTC)

Can somebody who knows about the subject re-write the following sentence from the intro?

  • In order to translate the probability of data given a hypothesis to the probability of a given hypothesis given the data it is necessary to use prior probability and background information.

I have to confess that I find it a bit hard to follow. Perhaps a few less "givens" might help. --BobSpring is sprung! 16:21, 19 July 2011 (UTC)

The second "given" isn't needed, but generally that sentence is about as good as it can get although it won't make much sense until you read "Probability of the hypothesis versus the data" below. ADK...I'll spit your petroglyph! 16:39, 19 July 2011 (UTC)
Still trying to get my brain around it. Is "translate" the best word? Would "express" be better? Or "convert" now I read it again? Or do I remain confused? Or is this special terminology?--BobSpring is sprung! 17:52, 19 July 2011 (UTC)

Bayesianism[edit]

Is that a word?--User:Brxbrx/sig 01:42, 26 October 2011 (UTC)

So a quick google search returns results for Bayesian, but not Bayesianism. It just seems really awkward to say Bayesian all the time. And how do I pronounce it, anyways? I'm not learning that funky phonetic speak, just spell it like you hear it if you please, thank you. RationalWiki is somewhat meant to educate people, right? I've got some learns to learn. Learn me up if you have the time.--User:Brxbrx/sig 19:11, 30 October 2011 (UTC)
bump --User:Brxbrx/sig 15:53, 31 October 2011 (UTC)
Bay-zee-anne. Link=User:Armondikov...I'll complement your blimp! 15:02, 1 November 2011 (UTC)
Thanks. But is Bayesianism a word?--User:Brxbrx/sig 15:32, 1 November 2011 (UTC)
Yes. Nutty Roux100x100 anarchy symbol.svg 17:42, 1 November 2011 (UTC)

Bayes jokes[edit]

If you're that sort of person. Scarlet A.pngsshole 18:12, 23 January 2012 (UTC)

Almost painfully bad enough to be genius...almost. Nebuchadnezzar (talk) 18:20, 23 January 2012 (UTC)

Bayes' equation[edit]

Shouldn't the denominator have a index for different hypotheses? I've been trying to add one, but I can't get the format right.Fdof (talk) 02:40, 11 August 2012 (UTC)

Listed Equation[edit]

What's up with the sigma in the listed equation? I don't understand its meaning in this context.— Unsigned, by: 184.153.153.175 / talk / contribs

It's a summation, though it appears the author of that section is using some weird form of Einstein notation such that the summation is over repeated indices (in this case h). - Grant (talk) 03:43, 8 April 2014 (UTC)
I've explicitly added the summation index to make it more clear. - Grant (talk) 03:46, 8 April 2014 (UTC)
Thanks! Wouldn't it be simpler to just indicate the sum with a plus sign? It's not going to take up that much more space and it's clearer to more people... I, for one, probably wouldn't be able to deduce how the index works here without googling if I didn't already know Bayes' Theorem (I've never seen sigma without an explicit range before; I don't really know stats). 169.244.25.226 (talk) 17:11, 8 April 2014 (UTC)
It's common shorthand to omit the explicit range when it can easily be inferred. In this case, the inference is that the range is over all possible hypotheses, and thus doesn't have a determined endpoint. This is common in all fields that frequently use sums, including physics, most fields of mathematics, and so on. In many cases, a sum like this can be written with a range, say, i = (1,2,3,...,n), where n is the number of hypotheses, but I'm not sure adding that explicitly would clear things up. Using a plus sign also isn't ideal because the whole idea is that the summation is over all possible hypotheses, not just one or two. I'm not sure how one could express that without using proper summation notation (indices included). - Grant (talk) 19:36, 8 April 2014 (UTC)
Ah, I hadn't considered application of the theorem to more than two competing hypotheses... fair enough.— Unsigned, by: 169.244.25.226 / talk / contribs

Odds Formula[edit]

There is an alternative formulation of Bayes' equation in terms of odds (i.e. 1:100), which is often easier to calculate with. When expressed with odds, the equation becomes PO = IO * P(d|h)/P(d|~h). I tried to add it to the article, but I kept getting a syntax error with the math tags. When multiplying the odds by the strength factor (the probability ratio), you only multiply one side of the odds (usually the left side). EyeSky (talk) 22:27, 18 February 2021 (UTC)