Oh shit, it's the math threadPost your archive of very detailed graphs here The grounding-topic for this instance of 「Math Thread」 is: the application of mathematics to real-world problemsTranslator's note: Statistics is mathematics as well; I think past 「fuck

Ah, here's an interesting, short read for you all about the Königsberg bridge problemhttps://nrich.maths.org/2484Pretty much, is it possible to walk across all the bridges in the figure exactly once without doubling back across one of them?[spoiler]The answer is no, and the proof is very interesting![/spoiler]This problem led the // led to much interest in developing graph theory!

Here's a neat application I'm working onThe set-up is this:Every picture I own has some content in it, which can be "tagged"These tags are things like "blue sky" or "woman wth parasol", and can be used to describe a pictureIn a much more practical sense, it is evident that almost every picture tagged "akemi homura" also has the "puella magi madoka magica" tagSimilarly, every picture with the "Mizunashi Akari" tag has the "Aria" tagBut the trick is that not every picture is already tagged; mostly because not every picture has been tagged yet. It also happens that in these cases the correlation of "Mizunashi Akari" with "Aria" is damn near 1, but there are other cases where the correlation is lower with other tags. In short:What is a good metric to tell if two tags are "well-related"? Is it a correlation, or is there a better metric to tell? It's my understanding that correlation is a computationally expensive statistic to derive.I will post the dataset here soon so you can glance through it

>>4694

>>4694 coolness!

>>4694Why do you want to correlate these things?Heres r

>>4703I want to make inferences about how much ass a show has w/o having to power through itActually, I'm interested in esoterically drawing connections between tags but I'm no statistics-boyOn second thought, the correlation coeff doesn't seem too complex to compute; if it's taken that the sample space is all indexed pictures, I only need to iterate through the list once for every tag

>>4707Six degrees of Hitler, but with dragon ball z tiddiesniceAre you wanting to correlate everything with everything else?Asuka <==> ass <==> monster monsume <==> tits <==> bulma (milf) <==> dragon ball z (.8) //_(.8)___(.6)_______________(.95)___(.6)___________(1)______FuckI hope you understand my idea here.

Hello friends, I bring an interesting datasethttps://bm.howler.space/trash/hooya.txtThis is ofc a tree-style diagram so that it's easy to read off valuesI think it goes w/o saying that the parent node of each branch is a particular case, and then the children of each branch are how many times the child node occurs in unison (on the same picture) as the parentIf you're interested in a more raw data format, here is a .csv file which you can import into Excel or somethinghttps://bm.howler.space/trash/hooya_raw.csvThe columns are listed at the top of the file, but for posterity they are:TagX, which corresponds to a parent tag in the tree diagram I posted aboveX Frequency, which is simply how many pictures are tagged as TagXTagY, which corresponds to a child tag in the tree diagramY Frequency given X, which is the number of times the tag TagY occurs in files which are tagged with TagXP(Y|X), a simple calculation of ("Y Frequency given X"/"X Frequency")I posit that P(Y|X) is a good metric which is simple to calculate, and can lead to some insight into the relevance of TagY with TagXI'm working on some other projects ATM so the implications of this dataset will have to wait for a little bit; I hope you find this data interesting, \math\!

moar data fun herehttps://bm.howler.space/hub/hooYa/stats/index.php?tagsclick on a tag for more detail than you ever fucking asked forsimple man's summary here:https://bm.howler.space/hub/hooYa/stats/index.php?overviewyou can help us to organize our pictures by adding appropos tags athttps://bm.howler.space/hub/hooYa/random.phpwhich is a randomly selected vein of the pictures we have yet to organizeand if you create an account (https://bm.howler.space/hub/common/util/acc_create.php) + sign in, you can fight on the high score table at https://bm.howler.space/hub/common/util/users.php//fucking ^ https://bm.howler.space/hub/common/util/acc_create.php

What interesting things would happen if there were 2 extra days between every day of the week? How many days would there be in a week, or would the week expand to exist forever?>>51277 days is pretty rough7 + (2 * 6) days would be worsean extra weeknd would be welcome, but maybe it is the case that we have an extra weeknd in the middle of every two weeks

Don't be so depressive. The week is long enough, I can barely stomach it as it is.

>>5126If I had 66% more day per day, id be stoked.

Days are too long as they are half the time

if i had a pizza for every time i saw a bad excel graph at a non-STEM presentation, id have a lot of pizzas

>>5286aoe2 made those charts feel very natural

how do i differentiate my 1 subscripts from my commas

>>5421impossible; syntacticaly seperate your ideas with 1 subscripts instead of commas for consistency,,,,,,,,,,,,,

Powered by Wakaba Formerly powered by Doushio Maintained by Prettyboy Yumi