CODEWITHSUNDEEP
rplotmosaic-plot
user12635841
user12635841

Reputation:

How to plot a mosaicplot for a 1x1 contingency table?

Plotting a 1-by-1 contingency table returns an error:

dat <- read.table(textConnection('
foo bar
TRUE TRUE
TRUE TRUE
'), header = TRUE, colClasses=c('logical', 'logical'))
mosaicplot(table(dat))

Error in rep.int(0, ydim) : invalid 'times' value

As I learned, the code in the mosaicplot function doesn't allow to plot a 1-by-1-table. But then, how do I plot a mosaicplot of that table?

Background.

I am plotting a series of dynamically created tables, some of which sometimes happen to have only one column and one row, at other times they have more dimensions. Having an undivided rectangle in that series of mosaicplots is valuable information and easily grasped in that visual representation.

Upvotes: 2

Views: 180

Answers (2)

Henrik
Henrik

Reputation: 67778

One possibility is to coerce the variables to be plotted to factor and specify the possible outcomes in levels (in the desired order). Then zero count cells will be represented as thin lines.

dat[] = lapply(dat, factor, levels = c(TRUE, FALSE))
mosaicplot(table(dat))

enter image description here

Upvotes: 3

jay.sf
jay.sf

Reputation: 73272

If you don't want the thin lines of @Henrik's (great!) solution you could substitute one-dimensional cases with a custom plot function.

mpOneDim <- function(tbl, title="table(dat)") {
  dn <- attr(tbl, "dimnames")
  labs <- names(dn)
  levels <- unlist(dn)
  plot.new()
  rect(0.0125, -.035, .985, .99, col="gray", border=1)
  mtext(title, line=1.55, font=2, cex=1.2)
  mtext(labs[1], 1, 1)
  mtext(labs[2], 2, 1)
  mtext(levels[2], 2, -1.15, cex=.7)
  mtext(levels[2], 3, -.75, cex=.7)
}
mpOneDim(table(dat))

enter image description here

In your function do something like this:

if (sum(dim(table(dat))) <= 2) {
  mpOneDim(table(dat))
} else {
  mosaicplot(table(dat))
}

Data:

dat <- structure(list(foo = c(TRUE, TRUE), bar = c(TRUE, TRUE)), class = "data.frame", row.names = c(NA, 
-2L))

Upvotes: 0

Related Questions