R help - Dec 2008 - Hmisc Dotplot with confidence intervals and panel.points problem

Hello useRs,
I have a question regarding the function Dotplot from the Hmisc package: I want
two things:
1) confidence intervals around the dots
2) some additional "annotation" points plotted in the graphic

I can easily achieve (1) by constructing an appropriate object with Cbind. But
for (2) when I use the panel=function argument the confidence intervals of the
original plot are gone.

This is my (simplified) code:

Dotplot(
  resultrow ~ Cbind(ESTIM, L95, U95),
  data=results,
  abline=list(v=0),
  scales=list(
    y=list(
      labels=paste(as.character(foo$bar),
      at=results$nranalysis,
      cex=0.5
    )
  ),
  xlab="some label",
  ylab="",
  xlim=c(-10,10),
  main=paste("a title"),
  subscripts = T,
  panel=function(...)
  { #panel.xYplot(...)
    panel.points(x=rep(9, length(toobig)), y=results$noanalysis[toobig], pch=62)
    panel.dotplot(...)
  }
)
I tried a number modifications of this code but without success.
I'm sure I'm doing something wrong here, but I cannot figure it out...
any ideas?

Thanks in advance!
Peter
















Background information:
======================My problem is that some of the values to be plotted have a
big range, so that one cannot see what is happening near zero (which is of more
interest for me), because the x-axis is properly scaled by R. So I use xlim to
restrict drawing to -10 to 10. But then some points and confidence intervals are
no more visible. So I decided to put a character ">" at x-value 9
for the ones lying above 10 and the code above leads to the correct plotting of
the original data and markers in the right places but the confidence intervals
are gone. I guess that the confidence intervals are themselves plotted by a
panel function which in some way gets overwritten by my custom one.


System Information:
==================> sessionInfo()
R version 2.7.2 (2008-08-25)
i386-pc-mingw32

locale:
LC_COLLATE=German_Germany.1252;LC_CTYPE=German_Germany.1252;LC_MONETARY=German_Germany.1252;LC_NUMERIC=C;LC_TIME=German_Germany.1252

attached base packages:
[1] stats     graphics  grDevices datasets  tcltk     utils     methods  
[8] base     

other attached packages:
[1] svSocket_0.9-5 TinnR_1.0.2    R2HTML_1.59    Hmisc_3.4-4   

loaded via a namespace (and not attached):
[1] cluster_1.11.11 grid_2.7.2      lattice_0.17-17 svMisc_0.9-5   
[5] tools_2.7.2

LOVE at LYCOS ist jetzt ganz neu ? melde dich kostenlos an, lerne neue Leute
kennen und finde deinen Flirt! http://love.lycos.de/start.action

PSJ wrote:
> Hello useRs,
> I have a question regarding the function Dotplot from the Hmisc package: I
want two things:
> 1) confidence intervals around the dots
> 2) some additional "annotation" points plotted in the graphic
> 
> I can easily achieve (1) by constructing an appropriate object with Cbind.
But for (2) when I use the panel=function argument the confidence intervals of
the original plot are gone.
> 
> This is my (simplified) code:
> 
> Dotplot(
>   resultrow ~ Cbind(ESTIM, L95, U95),
>   data=results,
>   abline=list(v=0),
>   scales=list(
>     y=list(
>       labels=paste(as.character(foo$bar),
>       at=results$nranalysis,
>       cex=0.5
>     )
>   ),
>   xlab="some label",
>   ylab="",
>   xlim=c(-10,10),
>   main=paste("a title"),
>   subscripts = T,
>   panel=function(...)
>   { #panel.xYplot(...)
>     panel.points(x=rep(9, length(toobig)), y=results$noanalysis[toobig],
pch=62)
>     panel.dotplot(...)
>   }
> )
> I tried a number modifications of this code but without success.
> I'm sure I'm doing something wrong here, but I cannot figure it
out... any ideas?
> 
> Thanks in advance!
> Peter
Peter,

The panel function must run panel.Dotplot(...) and not panel.dotplot. 
Inside panel you can run whatever else you need.

Frank
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> Background information:
> ======================> My problem is that some of the values to be
plotted have a big range, so that one cannot see what is happening near zero
(which is of more interest for me), because the x-axis is properly scaled by R.
So I use xlim to restrict drawing to -10 to 10. But then some points and
confidence intervals are no more visible. So I decided to put a character
">" at x-value 9 for the ones lying above 10 and the code above
leads to the correct plotting of the original data and markers in the right
places but the confidence intervals are gone. I guess that the confidence
intervals are themselves plotted by a panel function which in some way gets
overwritten by my custom one.
> 
> 
> System Information:
> ==================>> sessionInfo()
> R version 2.7.2 (2008-08-25)
> i386-pc-mingw32
> 
> locale:
>
LC_COLLATE=German_Germany.1252;LC_CTYPE=German_Germany.1252;LC_MONETARY=German_Germany.1252;LC_NUMERIC=C;LC_TIME=German_Germany.1252
> 
> attached base packages:
> [1] stats     graphics  grDevices datasets  tcltk     utils     methods  
> [8] base     
> 
> other attached packages:
> [1] svSocket_0.9-5 TinnR_1.0.2    R2HTML_1.59    Hmisc_3.4-4   
> 
> loaded via a namespace (and not attached):
> [1] cluster_1.11.11 grid_2.7.2      lattice_0.17-17 svMisc_0.9-5   
> [5] tools_2.7.2
> 
> LOVE at LYCOS ist jetzt ganz neu ? melde dich kostenlos an, lerne neue
Leute kennen und finde deinen Flirt! http://love.lycos.de/start.action
> 
> 
> ------------------------------------------------------------------------
> 
> ______________________________________________
> 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.

-- 
Frank E Harrell Jr   Professor and Chair           School of Medicine
                      Department of Biostatistics   Vanderbilt University

Dear Frank,
thanks for your fast answer...
> The panel function must run panel.Dotplot(...) and not panel.dotplot. 
> Inside panel you can run whatever else you need.
> 
> Frank
thanks a lot, this made it work!

Cheers,
Peter