[R] error bars in lattice xyplot *with groups*

[Manuel Morales]

Hi Mike,

If you're not committed to using a panel function, another option is to
use the function lineplot.CI, available in the package sciplot at
http://mutualism.williams.edu/sciplot

E.g.
# Define and generate variables in "long" format
range <- vector()
voice <- vector()

string <- strsplit(as.character(singer$voice.part)," ")
for(i in 1:dim(singer)[1]) {
  range[i] <- string[[i]][1] 
  voice[i] <- string[[i]][2]
}

# Define function for CI
conf.int <- function(x) {
              st <- boxplot.stats(x)
              c((st$conf[2]-st$conf[1])/2)
              }

# Plot
library(sciplot)
lineplot.CI(response=height, x.factor=voice, trace.factor=range,
            data=singer, fun=median, ci.fun=conf.int)

lineplot.CI(response=height, x.factor=voice.part, data=singer,
            fun=median, ci.fun=conf.int)


Manuel


On Tue, 2006-06-06 at 00:20 -0300, Mike Lawrence wrote:
> Hi all,
> 
> I'm trying to plot error bars in a lattice plot generated with xyplot. Deepayan
> Sarkar has provided a very useful solution for simple circumstances
> (https://stat.ethz.ch/pipermail/r-help/2005-October/081571.html), yet I am
> having trouble getting it to work when the "groups" setting is enabled in
> xyplot (i.e. multiple lines). To illustrate this, consider the singer data
> generated by the above linked solution previously submitted:
> 
> #####################
> library(lattice)
> singer.split <-
>     with(singer,
>          split(height, voice.part))
> 
> singer.ucl <-
>     sapply(singer.split,
>            function(x) {
>                st <- boxplot.stats(x)
>                c(st$stats[3], st$conf)
>            })
> 
> singer.ucl <- as.data.frame(t(singer.ucl))
> names(singer.ucl) <- c("median", "lower", "upper")
> singer.ucl$voice.part <-
>     factor(rownames(singer.ucl),
>            levels = rownames(singer.ucl))
> 
> #now let's split up the voice.part factor into two factors,
> singer.ucl$voice=factor(rep(c(1,2),4))
> singer.ucl$range=factor(rep(c("Bass","Tenor","Alto","Soprano"),each=2))
> 
> #here's Deepayan's previous solution, slightly modified to depict
> #  the dependent variable (median) and the error bars on the y-axis
> #  and the independent variable (voice.part) on the x-axis
> prepanel.ci <- function(x, y, ly, uy, subscripts, ...)
> {
>     x <- as.numeric(x)
>     ly <- as.numeric(ly[subscripts])
>     uy <- as.numeric(uy[subscripts])
>     list(ylim = range(y, uy, ly, finite = TRUE))
> }
> panel.ci <- function(x, y, ly, uy, subscripts, pch = 16, ...)
> {
>     x <- as.numeric(x)
>     y <- as.numeric(y)
>     ly <- as.numeric(ly[subscripts])
>     uy <- as.numeric(uy[subscripts])
>     panel.arrows(x, ly, x, uy, col = "black",
>                  length = 0.25, unit = "native",
>                  angle = 90, code = 3)
>     panel.xyplot(x, y, pch = pch, ...)
> }
> 
> 
> #this graph works
> xyplot(median ~ voice.part,
> 	data=singer.ucl,
> 	ly = singer.ucl$lower,
> 	uy = singer.ucl$upper,
> 	prepanel = prepanel.ci,
> 	panel = panel.ci,
> 	type="b"
> )
> 
> #this one does not (it will plot, but will not seperate the groups)
> xyplot(median ~ voice,
> 	groups=range,
> 	data=singer.ucl,
> 	ly = singer.ucl$lower,
> 	uy = singer.ucl$upper,
> 	prepanel = prepanel.ci,
> 	panel = panel.ci,
> 	type="b"
> )
> 
> ####################################
> 
> Any suggestions?
>