[R] FIX: pscales=list(...) and splom

Jens Scheidtmann JensScheidtmann at web.de
Sat Mar 8 15:31:48 CET 2003

Dear R Users,

When plotting with "splom" I tried to use the pscales=list(...) feature
Unfortunately it didn't work at all. Instead the scales always were

So I looked at the source of panel.pairs and found:

"draw <- is.numeric(pscales) && pscales != 0"

which rather has to be:

"draw <- is.list(pscales) || (is.numeric(pscales) && pscales != 0)"

With this change, the axis labels could be generated from the
supplied ...$at and ...$lab components of the pscales entries.
(BTW: the documentation says the components should be $at and $labels. This
is wrong!)


* The limits of the axes are still generated from the data.

To make panel.pairs honor the ranges supplied in the pscales list, one has
to change

  if (n.var > 0) {
    lim <- list(1:n.var)
    for (i in 1:n.var) {
      lim[[i]] <- extend.limits(range(as.numeric(z[, i]),
                                      na.rm = TRUE))


  if (n.var > 0) {
    lim <- list(1:n.var)
    for (i in 1:n.var) {
      lim[[i]] <- extend.limits(range(as.numeric(z[, i]),
                                      na.rm = TRUE))
      if (is.list(pscales)) {
        lim[[i]] <- extend.limits(range(pscales[[i]]$at))

* The labels are still not as expected.

To fix these, occurences of "as.character(axls ..." have to be replaced by
"labels ..."
(see enclosed diff).

Now a call like the following works and displays
the correct ranges and labels:

# Let axes start by 0 and use other decimal separator
pscal <- list(1:4)
pscal[[1]] <- list(at=seq(0,10,2), lab=paste(seq(0,10,2)))
pscal[[2]] <- list(at=seq(0,5,1), lab=paste(seq(0,5,1)))
pscal[[3]] <- list(at=seq(0,8,2), lab=paste(seq(0,8,2)))
pscal[[4]] <- list(at=seq(0,3,0.5),
splom(~iris[1:4], groups = Species, data = iris,
      panel = panel.superpose,
      key = list(title = "Three Varieties of Iris",
        columns = 3,
        points = list(pch = super.sym$pch[1:3],
          col = super.sym$col[1:3]),
        text = list(c("Setosa", "Versicolor", "Virginica"))),

Please CC me on your replies, because I am not subscribed to this list.



Jens Scheidtmann
-------------- next part --------------
A non-text attachment was scrubbed...
Name: panel.pairs.diff
Type: application/octet-stream
Size: 6777 bytes
Desc: not available
Url : https://stat.ethz.ch/pipermail/r-help/attachments/20030308/5bbf0a63/panel.pairs.obj

More information about the R-help mailing list