On 10/27/06, Stefan Grosse <[EMAIL PROTECTED]> wrote:
> I have the following Data structure
>
> $ step45     : Factor w/ 2 levels
> $ obserror   : num  6.2 6.2 5.6 6.6 6.6 ...
> $ Mon        : num  2.2 2.0 1.0 3.2 2.0 ...
> $ inc.comp   : num  4 5 2 5 5 5 5 5 4 4 ...
>
> all I wanted to do is plotting Mon against obserror, the colors should
> be by step45 and the size of the symbol should be according to inc.comp
> so I did this:
>
> qplot(obserror,Mon,data=obscomp,col=inc.comp,col=step45)

I think you mean
> qplot(obserror,Mon,data=obscomp, size=inc.comp,col=step45)


> unfortunately the size of the is something I do not want it to be, the
> legend for inc.comp says: 4, 2.25, 1 ,0.25 , 0 I suppose this are weights?

I think there are two things here that are confusing you (both my fault)

 * the size aesthetic actually modifies the area (ie. sqrt(size)) of
the points, because that is better perceptually

 * the legend isn't very good

But it is mapping the inc.comp to the size of the variable.


> I needed the as.factor otherwise the plot is again different. But that
> did not work with size. OK I thought this is a second best solution but
> unfortunately the symbols are to small. But setting a size = 3 which I
> thout adds a constant size everywhere (somehow expectedly) did add a
> third legend "3" with 3.61, 3.42, ...

qplot currently doesn't have any way to set an aesthetic to a fixed
value (it can't tell that you're specifying an aesthetic value not a
data value).  You can do this in two steps though

p <- qplot(wt, mpg, data=mtcars, type=NULL, shape=cyl)
ggpoint(p, size=3)

> Another thing what I like to complain about is that qplot is using only
> half of the windows graphics device space (especially wen adding this
> size =3) which is not only ugly but a waste of space...

That sounds like a bug.  Could you please provide a reproducible example?

Thanks,

Hadley

______________________________________________
[email protected] 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.

Reply via email to