Re: [R] partial R

From: John Fox <jfox_at_mcmaster.ca>
Date: Mon 02 Apr 2007 - 15:54:10 GMT


Dear Pedram,

I think that you're confusing component+residual ("partial-residual") plots with added-variable ("partial-regression") plots. The latter shows residuals for Y and a particular X regressed on all the other X's, and the correlation in an added-variable plot is therefore the partial correlation between a particular X and Y "controlling for" all the other X's. This is exactly what the partial.cor() function computes. If I follow correctly what you want, you just have to square the partial correlations.

Regards,
 John



John Fox
Department of Sociology
McMaster University
Hamilton, Ontario
Canada L8S 4M4
905-525-9140x23604
http://socserv.mcmaster.ca/jfox

> -----Original Message-----
> From: Pedram Rowhani [mailto:rowhani@geog.ucl.ac.be]
> Sent: Monday, April 02, 2007 11:06 AM
> To: John Fox; 'Michael Kubovy'
> Cc: r-help@stat.math.ethz.ch
> Subject: RE: [R] partial R
>
> Dear John and Michael,
>
> sorry for the poor explanation of my question.
>
> What I am looking for is the partial R-squares, to estimate
> the proportion of unexplained variation of y that becomes
> explained with the addition of variable x cr.plots does give
> me the plots but I am looking for a way to get the actual
> partial R squares, which would correspond to the R squares of
> those plots
>
> I hope that things are a bit more clear now
>
> I guess I can calculate the SSE's and SSR's of the different
> models myself and then use the formula for partial R square
> but things would be just simpler if there was already a
> function implemented in R
>
> Thanks again
> Pedram
>
> At 16:20 02/04/2007, John Fox wrote:
>
>
> Dear Michael and Pedram,
>
> I'm afraid that Pedram's question is unclear, since the
> subject line refers
> to "partial R" (which might have been intended as
> "partial r," or "partial
> correlation"), while the message itself refers to
> "partial regression
> coefficients."
>
> The latter are simply the coefficients returned by
> lm(); there is a
> partial.cor() function in the Rcmdr package that
> computes partial
> correlations. It's so simple, that I'll just reproduce it here:
>
> partial.cor <-
> function (X, ...)
> {
> R <- cor(X, ...)
> RI <- solve(R)
> D <- 1/sqrt(diag(RI))
> R <- -RI * (D %o% D)
> diag(R) <- 0
> rownames(R) <- colnames(R) <- colnames(X)
> R
> }
>
> cr.plots() in the car package produces partial-residual plots
> ("component+residual plots") but returns neither
> partial correlations nor
> partial-regression coefficients.
>
> Regards,
> John
>
> --------------------------------
> John Fox
> Department of Sociology
> McMaster University
> Hamilton, Ontario
> Canada L8S 4M4
> 905-525-9140x23604
> http://socserv.mcmaster.ca/jfox
> <http://socserv.mcmaster.ca/jfox>
> --------------------------------
>
> > -----Original Message-----
> > From: r-help-bounces@stat.math.ethz.ch
> > [ mailto:r-help-bounces@stat.math.ethz.ch
> <mailto:r-help-bounces@stat.math.ethz.ch> ] On Behalf Of
> Michael Kubovy
> > Sent: Monday, April 02, 2007 9:16 AM
> > To: Pedram Rowhani
> > Cc: r-help@stat.math.ethz.ch
> > Subject: Re: [R] partial R
> >
> >
> > On Apr 2, 2007, at 5:49 AM, Pedram Rowhani wrote:
> >
> > > i am wondering if there is a command in R that will
> give me the
> > > partial regression coefficients
> >
> > To answer your question, you could have started with
> > RSiteSearch('partial regression')
> >
> > It's then likely that you would figured out that one
> way to proceed is
> > install.packages('car')
> > ?cr.plots
> >
> > (You may have to restart R to get the help on a
> newly-installed
> > package.)
> > _____________________________
> > Professor Michael Kubovy
> > University of Virginia
> > Department of Psychology
> > USPS: P.O.Box 400400 Charlottesville, VA 22904-4400
> > Parcels: Room 102 Gilmer Hall
> > McCormick Road Charlottesville, VA 22903
> > Office: B011 +1-434-982-4729
> > Lab: B019 +1-434-982-4751
> > Fax: +1-434-982-4766
> > WWW: http://www.people.virginia.edu/~mk9y/
> <http://www.people.virginia.edu/~mk9y/>
> >
> > ______________________________________________
> > R-help@stat.math.ethz.ch mailing list
> > https://stat.ethz.ch/mailman/listinfo/r-help
> <https://stat.ethz.ch/mailman/listinfo/r-help>
> > PLEASE do read the posting guide
> > http://www.R-project.org/posting-guide.html
> <http://www.r-project.org/posting-guide.html>
> > and provide commented, minimal, self-contained,
> reproducible code.
> >
>
> Pedram Rowhani Ardekani
> University of Louvain
> Department of geography
> Place Louis Pasteur, 3
> 1348 Louvain-la-Neuve (Belgium)
> tel (+32) 010/47.28.70
> fax (+32) 010/47.28.77
> e-mail rowhani@geog.ucl.ac.be
> http://www.geo.ucl.ac.be/Recherche/Teledetection/index.html
> <http://www.geo.ucl.ac.be/Recherche/Teledetection/index.html>
>



R-help@stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code. Received on Tue Apr 03 01:58:54 2007

Archive maintained by Robert King, hosted by the discipline of statistics at the University of Newcastle, Australia.
Archive generated by hypermail 2.1.8, at Mon 02 Apr 2007 - 16:30:37 GMT.

Mailing list information is available at https://stat.ethz.ch/mailman/listinfo/r-help. Please read the posting guide before posting to the list.