turn on suggestions

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

Find a Community

- Home
- /
- Analytics
- /
- Stat Procs
- /
- Proc cancor

Topic Options

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

01-29-2015 08:50 AM

Hi All -

In canonical correlation, the ith eigenvalue of the matrix R (where R is the product of the inverse of the variance-covariance matrix of y, the covariance matrix of x and y, the inverse of the variance-covariance matrix of x, and the covariance matrix of y and x) is equal to the ith canonical correlation squared.

Why then does SAS output the eigenvalues derived by CanRsq / (1 - CanRsq) rather than having the eigenvalues be equal to CanRsq?

What is the significance behind eigen values? Also It would be juicy if someone of you tell me with simple example when '**proc concor'** should be used over '**proc corr'**? B ecause I'm not clear although I tried to difference between them.

Thanks!

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Posted in reply to Babloo

01-30-2015 12:43 AM

Any helpful answers?

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Posted in reply to Babloo

01-30-2015 07:57 AM

Why then does SAS output the eigenvalues derived by CanRsq / (1 - CanRsq) rather than having the eigenvalues be equal to CanRsq?

The documentation for PROC CANCORR is pretty clear on this point. It says: "Eigenvalues of INV(E)*H, which are equal to CanRsq/(1–CanRsq), where CanRsq is the corresponding squared canonical correlation."

But you can convert CanRsq/(1-CanRsq) into CanRsq if you want.

Also It would be juicy if someone of you tell me with simple example when '

proc concor'should be used over 'proc corr'?

PROC CORR gives you all requested pairwise correlations (the correlation of, for example, X1 with X2)

PROC CANCORR is a multivariate procedure, it gives you correlations of linear combinations of variables in the VAR statement with linear combinations of variables in the WITH statement.

--

Paige Miller

Paige Miller

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Posted in reply to PaigeMiller

01-30-2015 10:54 AM

Thanks for your reply.

Do we need to concentrate on the values of CanRsq/(1–CanRsq) when there is no correlation and also no significant values?

If you can explain the eigen value calculation (INV(E)*H) it would be great.

Thanks again.

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Posted in reply to Babloo

01-30-2015 11:10 AM

Do we need to concentrate on the values of CanRsq/(1–CanRsq) when there is no correlation and also no significant values?

Probably not

If you can explain the eigen value calculation (INV(E)*H) it would be great.

Explain how eigenvalues are calculated? No, I can't do that.

Or explain why they are calculated? They are calculated to determine the relative proportion of variability explained by each canonical component.

--

Paige Miller

Paige Miller