Talk:Covariance and correlation

Stub question

This page is not really a stub, it is pretty complete and concise explanation of differences between the two concepts. If you edit this page, please do so sparingly. —Preceding unsigned comment added by 137.100.46.254 (talk) 18:31, 14 January 2010 (UTC)

No doubt there are many reasons why this has been put down as a stub. One of them is the lack of citations, both the relations stated and more importantly for the other statements of supposed "fact". Melcombe (talk) 10:12, 18 January 2010 (UTC)

How are people supposed to discover this article? It's not linked from either Covariance or Correlation, and those articles don't discuss the relation between both concepts in a prominent place.--88.73.53.132 (talk) 16:14, 1 November 2011 (UTC)

WTF?

"The covariance of a variable with itself (i.e. X = Y) is called the variance" You can't tell me that sentence makes a damn bit of sense. It also contradicts: http://en.wikipedia.org/wiki/Variance, which says "The variance is a measure of how far a set of numbers is spread out". Theblindsage (talk) 22:34, 27 February 2012 (UTC)

Thanks for your requests for clarification here and in the next two threads (although if you had been less strident someone might have taken you more seriously, read all the way through your comments, and clarified these things a long time ago).

Yes it makes sense, and is a standard sentence. I've put into the text that this arises when Y always takes on the same value as X. And no, it does not contradict the sentence from the other article. Loraof (talk) 17:37, 6 January 2017 (UTC)

WTF#2

"The correlation of a variable with itself is always 1" a) How can something correlate with itself? Ah, wait--it is just terrible grammar. "...(except in the degenerate case where the two variances are zero, in which case the correlation does not exist)" b) What is a degenerate case? At least give either an example or a link to what you are talking about. c) How can a variance be zero? d) Correlation does not exist, or is equal to zero? (Very different, those two). Theblindsage (talk) 22:34, 27 February 2012 (UTC)

(a) Nope, it's not terrible grammar (nor is it terrible wording, which is surely what you meant but is not the same thing). As with the covariance of a variable with itself, this occurs when Y always takes on the same value as X.

(b) Degenerate case is now wikilinked.

(c) I've clarified that the variance is zero if X always takes on the same single value.

(d) It is correct -- it does not exist since its computation would involve division by 0.

Loraof (talk) 17:37, 6 January 2017 (UTC)

WTF#3

The mean variables given above are not used. Perhaps the expected value of X should be replaced by the mean of X, and similarly for Y?

Expected value and mean (in its most common use) are synonyms. They are now linked and their relationship clarified in the text. Loraof (talk) 17:37, 6 January 2017 (UTC)

Huh?

The first section says that covariance = foo

Then it says that correlation = foo / bar

Then it says that covariance = correlation / bar

I am far from an expert in these matters, but even I can see that can't be right.

https://en.wikipedia.org/wiki/Correlation_and_dependence says that correlation is covariance divided by the sigmas so I've edited the page accordingly.

— Preceding unsigned comment added by 137.17.63.8 (talk) 14:06, 5 March 2019 (UTC)