Alejandro Ria�o
> > I used to program in Gauss, that's why I still think
> > about matrixes and vectors when I'm doing my own
> > programs. My question is the following
> >
> > I want to store the correlation between 2 variables by
> > each pair of countries and then store this number(that
> > would be r(rho)) in a vector, so at the end I would
> > have a column with the correlation for each in pair of
> > countries as a position in this vector.
> >
> > Do you know how can i do this in STATA ???
Scott Merryman
> There is probably a better way to do this, but here is one method
>
> levels countries, local(levels)
> matrix A = 1
> foreach x of local levels {
> qui corr var1 var2 if rep == `x'
> scalar rho`x' = r(rho)
> matrix a`x' = rho`x'
> matrix A = A \ a`x'
>
> }
> matdelrc A, r(1)
>
Comment on Scott's posting
==========================
Scott shows how to build up a solution
more or less from first principles,
which is always worth seeing.
To help understand his posting, note
that -levels- is user-written and on SSC
and that -matdelrc- is also user-written
and is part of a package issued in STB-50.
-matdelrc- could be avoided by using -nullmat()-.
Here is a shorter version (assuming i.a.
that he was mixing examples):
levels countries, local(levels)
foreach x of local levels {
qui corr var1 var2 if countries == `x'
matrix A = nullmat(A) \ r(rho)
}
and you will probably want to follow
with something like
matrix rownames A = `levels'
Alternative approach
=====================
Check out -vectorof- and
-matrixof-. They are both in the
-matrixof- package on SSC. Example:
. vectorof corr headroom-gear, y(mpg) r(rho) matname(corr)
produces a vector of correlations between a set of
variables and -mpg-.
But you may still need to program
your own, depending on your data structure.
At worst, this would be two nested -foreach- loops.
Nick
[email protected]
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
