[R] Clusters of adjacent bars in barchart()

[Paul Murrell]

Hi


> Can I create a barchart in lattice with clusters of adjacent bars, which is
> possible with barplot() when beside=TRUE?  How can I
> accomplish the grouping in lattice that is done with
> inputting a matrix to barplot()?


(after a quick look at the code) I don't think barchart() will do this.
Deepayan (the lattice author) is away at the moment, but I have cc'ed
him in case I have missed something obvious.

In the meantime, if you're interested, below is a panel function that
might do what you want.  It's a bit of a draft really because it ignores
the box.ratio and horizontal arguments.  Furthermore, it relies on the
current way that a matrix x argument gets passed to the panel function,
which is more by accident than design at this stage I think -- the
important point is that this may change in the future.

Anyway, hope it's of some help.

Paul

### customised barchart panel function
mybarchartpanel <- function(x, y,
                            # Choose some vaguely rational default
colours
                            col=trellis.par.get("regions")$col[seq(1,
                              length(trellis.par.get("regions")$col),
                              length=ncol)],
                            ...) {
  yscale <- current.viewport()$yscale
  # For each x value
  for (i in unique(x)) {
    # Allocate a region for multiple side-by-side bars
    # The 1/(1.5*ny) is a rough calculation to make sure the
    # regions don't overlap
    push.viewport(viewport(x=unit(i, "native"),
                           width=unit(1/(1.5*ny), "npc"),
                           # Allocate a sub-region for each 
                           # individual side-by-side bar
                           layout=grid.layout(1, ncol)))
    yy <- y[x == i]
    for (j in 1:ncol) {
      # Draw each individual side-by-side bar
      push.viewport(viewport(layout.pos.col=j, yscale=yscale))
      grid.rect(y=unit(0, "native"),
                height=unit(yy[j], "native"),
                just=c("centre", "bottom"),
                gp=gpar(fill=col[j]))
      pop.viewport()
    }
    pop.viewport()
  }
}

# A simple example of a single barchart
# The data are not very interesting, but they're good
# for checking that it's working :)
# NOTE that I have to rep() the x values so that x is the same
# length as y
ncol <- 3
ny <- 8
y <- matrix(1:(ny*ncol), ncol=ncol)
x <- factor(rep(letters[1:ny], ncol))
barchart(y ~ x, panel=mybarchartpanel)

# A more complicated example with multiple panels
ng <- 4
y <- matrix(1:(ny*ng*ncol), ncol=ncol)
x <- factor(rep(letters[1:ny], ncol*ng))
g <- factor(rep(1:ng, rep(ny, ng)))
barchart(y ~ x | g, panel=mybarchartpanel)
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._