> "Simon, Steve, PhD" schrieb:
>
> I just wrote an S-plus function that computes the statistical measures described in
>
>
> > corr.reg
> function(x, y)
> {
> ey.x <- resid(lm(y ~ x))
> ex.y <- resid(lm(x ~ y))
> rde.y <- cor(abs(ex.y), abs(y - mean(y)))
> rde.x <- cor(abs(ey.x), abs(x - mean(x)))
> rde.y - rde.x
> }
>
> Even if you don't use S-plus, you should be able to follow this. Note that the syntax
> lm(y~x) fits a linear regression model with y as the dependent variable. The rest of
> the syntax should be obvious.
>
> I tried it with some random uniform variables as follows:
>
> > x1 <- runif(50)
> > x2 <- runif(50)
> > y <- x1+x2
> > corr.reg(x1,y)
> [1] -0.7029148
> > corr.reg(x2,y)
> [1] -0.6784307
> > corr.reg(x1,x2)
> [1] -0.02448412
Hi Steve,
I wonder why the coefficents have so high absolute values; if I
compute them, then in your case of corr.reg(x1,y) I get even with
ideal data a maximum absolute value of ~0.5 .
This value occurs, if in
> rde.y <- cor(abs(ex.y), abs(y - mean(y)))
I have z-value variables as parameters as well as in
> > y <- x1+x2
> > corr.reg(x1,y)
The other aspects of your implementation seem to be correct.
Uhmm, btw, are x1 and x2 uncorrelated when created like this in
S-Plus? If not, the absolute rde-value should be smaller again.
Regards -
Gottfried
--
Gottfried Helms
Univ. Kassel.
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
. http://jse.stat.ncsu.edu/ .
=================================================================
