[R] Produce graph that looks nice on screen and on paper

Deepayan Sarkar deepayan.sarkar at gmail.com
Sun Nov 11 17:54:11 CET 2007


On Nov 8, 2007 11:22 AM, Van Campenhout Bjorn
<bjorn.vancampenhout at ua.ac.be> wrote:

>  On 11/6/07, Van Campenhout Bjorn <bjorn.vancampenhout at ua.ac.be> wrote:
>  >> Hi all,
>  >>
>  >> I made a dotplot() with lattice, which comes out nice on the graphics
>  >> device.  I can save this as a eps using postscript() and include this in
>  >> a word document.  This prints nice, but does not look good on screen.
>  >> If I produce a pdf, it is nice on screen, but not on paper.  How can I
>  >> save a graph that looks nice on paper and on screen?
>
>  >Although you haven't said so, I presume your 'paper' output is
>  >produced by a black and white printer?
>  >
>  >You have to define a "theme" (a collection of graphical parameters)
>  >that is "nice" for both screen and paper. Once you come up with such a
>  >collection of parameters, ?trellis.par.set should tell you how to set
>  >them globally. If you want someone else to come up with such a list,
>  >you are probably out of luck (but see ?col.whitebg for a not-very-good
>  >approach).
>
>  >-Deepayan
>
>  Thanks for your reply.  However, I am unsure that this is a "theme" issue.
> The problem is that if I run for instance:
>
>  names <- rep( c( "country1" , "country2" , "country3" , "country4" ), each
> = 4 )
>  groups <- rep( 1:4 , times=4 )
>  score <- runif(16)*groups
>  dotplot(names~score, groups=groups, col=c("grey","grey","black","black"),
> pch=c(21,19,21,19),cex=1.5)
>
>  this looks nice in the R graphics device. But if I save or copy the graph
> as a windows metafile (as recommended on the list when one wants to include
> a graph in a Word document)

Recommended by whom? I was led to believe (I don't remember all the
reasons) that it is better to regenerate the graph than copy it; i.e.,
you should open a new device using

emf(file="foo.emf")

(or whatever the proper incantation is), re-execute your commands, and
then close the device with

dev.off()

Does this give better results?

-Deepayan

> the points become slightly deformed when I paste
> it into word. Is this normal?
>
>  Bjorn
>



More information about the R-help mailing list