[R] symbols(x,y, circles=sqrt(N)) with lattice xyplot

[Greg Snow]

The symbols function is base graphics which does not play nicely with lattice graphics (without extra work).

Here is one approximation that works with lattice:

x <- runif(10)
y <- rnorm(10)
z <- runif(10, 1,5)

library(lattice)
library(TeachingDemos) # for panel.my.symbols and friends
xyplot( y ~ x, panel=panel.my.symbols, symb=ms.polygon(250), inches=sqrt(z)/5 )

hope this helps,

-- 
Gregory (Greg) L. Snow Ph.D.
Statistical Data Center
Intermountain Healthcare
greg.snow at imail.org
801.408.8111


> -----Original Message-----
> From: r-help-bounces at r-project.org [mailto:r-help-bounces at r-
> project.org] On Behalf Of Jacob Wegelin
> Sent: Monday, September 14, 2009 4:55 PM
> To: r-help at stat.math.ethz.ch
> Subject: [R] symbols(x,y, circles=sqrt(N)) with lattice xyplot
> 
> How would I create the following plot using lattice?
> 
> symbols( combPsummary$pastRate, combPsummary$finRate,
> circles=sqrt(combPsummary$N) )
> 
> The idea is to plot finRate vs pastRate using circles whose areas are
> proportional to the number of people in each group.
> 
> The following attempt does not really work:
> 
>    xyplot(
>       finRate ~ pastRate
>       , data=combPsummary
>       , panel=function( x, y, ...) {
>          panel.xyplot( x,y, type="n" )
>          symbols( x , y, circles=sqrt(combPsummary$N))
>          }
>       )
> 
> It apparently first draws a rectangle, i.e., the lattice panel, and
> then inside that rectangle it draws the same thing that is drawn by
> the symbols() command outside of a panel function. That is, it draws
> another rectangle in addition to the desired circles, and it writes
> its own axis labels.
> 
> I didn't find "symbols" in the index of Deepayan's book. And I don't
> think the xyplot help file deals with this either? I don't think there
> is a "lsymbols" or "panel.symbols" function?
> 
> (Bill Cleveland points out that the human eye does not interpret area
> accurately, so that making the area of a circle proportional to sample
> size may be a flawed approach. Thus I'd also be interested in any
> comments on how one might best graphically display x, y, and sample
> size in a single plot.)
> 
> Thanks for any tips
> 
> Jake Wegelin
> 
> ______________________________________________
> R-help at r-project.org 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.