Skip to main content
W
WalterSchrabmai12-Amethyst
12-Amethyst
June 29, 2016
Question

Covariance vs. Covariance-Variance-Matrix

  • June 29, 2016
  • 1 reply
  • 2779 views

Hi

I have a question to the cvar() build-in Function. I know how it is programmed. but

in my Shee the correct covar for my covar-var-matrice is covar.

Kan anyone tell me the relationship between the cvar and my covar calculation?

Thanks

Walter

kvar_q.gif

1 reply

V
24-Ruby III
VladimirN24-Ruby III
24-Ruby III
June 29, 2016

Mathcad correctly calculate the covariance of two samples (see the theory here: Covariance - Wikipedia, the free encyclopedia ). What do you mean by the phrase "From Excel Sheet"?

Pic_2.png

Pic_1.png

W
WalterSchrabmai12-AmethystAuthor
12-Amethyst
June 29, 2016

yes MC do it fine. But I am a little bit convused with the Variance-Covariaqnce-Matrix. This Matrix is used to get the standard errors (the variances in the diagonal) of the parameters, which you get when you make a LS fit to a data set in Mathematica f.e.. And this for my example there are the covariances off diagonal which calculates as I have done with covar.

Could you open my eyes what the Covariances MC calculates and the wiki sheet is about do?

Thanks

I have done this paper here. where the Covariance of mb is so:

cov_mb.gif

A
AlvaroDíaz12-Amethyst
12-Amethyst
June 30, 2016

HI.

From the paper:

It became apparent that the students, having used different

textbook sources, were applying two methods to calculate sc

and these yielded different values. The purpose of this article is

to identify the source of this disagreement and to rectify the problem.

That's exactly was set in other of yours questions: Error propagation - How can help?

There, the teacher mix both methods, here, shows them separately.

Those two methods are: one, from mechanical engineering, more like brute force algorithm, which takes the abs values of the components of the gradient and sum them. Other, more from chemical engineerig, which make complicated assumptions about probabilistic happenings about measures. Remember that the entire probabilistic theory is a measure theory and the word "assign" is actually "measure". What the paper do is consider both, and use the "mech eng error measure" as a geometrical error measure, making more easy they maths.

Could you open my eyes what the Covariances MC calculates and the wiki sheet is about do?: MC calculate the statical one.

Well. There are some problem also. "True" statical could be for infinites samples. So, more correct: MC calculate the statical sampled one.

Best regards. Alvaro.

Terms & ConditionsCookie settingsAccessibility statement