How can I remove empty factors from ggplot2 facets?

I'm trying to modify an example of a simple forest plot by introducing facets according to a factor variable.

Assuming data of this structure:

test <- structure(list(characteristic = structure(c(1L, 2L, 3L, 1L, 2L
), .Label = c("Factor1", "Factor2", "Factor3"), class = "factor"), 
    es = c(1.2, 1.4, 1.6, 1.3, 1.5), ci_low = c(1.1, 1.3, 1.5, 
    1.2, 1.4), ci_upp = c(1.3, 1.5, 1.7, 1.4, 1.6), label = structure(c(1L, 
    3L, 5L, 2L, 4L), .Label = c("1.2 (1.1, 1.3)", "1.3 (1.2, 1.4)", 
    "1.4 (1.3, 1.5)", "1.5 (1.4, 1.6)", "1.6 (1.5, 1.7)"), class = "factor"), 
    set = structure(c(1L, 1L, 1L, 2L, 2L), .Label = c("H", "S"
    ), class = "factor")), .Names = c("characteristic", "es", 
"ci_low", "ci_upp", "label", "set"), class = "data.frame", row.names = c(NA, 
-5L))

And running the code:

p <- ggplot(test, aes(x=characteristic, y=es, ymin=ci_low, ymax=ci_upp)) + geom_pointrange() +
  coord_flip() + geom_hline(aes(x=0), lty=2) + 
  facet_wrap(~ set, ncol = 1) +
  theme_bw() + 
  opts(strip.text.x = theme_text())

Produces output like that:

All good so far. However, I'd like to get rid of empty Factor3 level from my lower panel and cannot find a way to do that. Is there any way to do that?

Thanks for help.

Solution 1:

EDIT Updated to ggplot2 0.9.3

Here's another solution. It uses facet_grid and space = "free"; also it uses geom_point() and geom_errorbarh(), and thus there is no need for coord.flip(). Also, the x-axis tick mark labels appear on the lower panel only. In the code below, the theme command is not essential - it is used to rotate the strip text to appear horizontally. Using the test dataframe from above, the following code should produce what you want:

library(ggplot2)

p <- ggplot(test, aes(y = characteristic, x = es, xmin = ci_low, xmax = ci_upp)) +
   geom_point() +
   geom_errorbarh(height = 0) +
   facet_grid(set ~ ., scales = "free", space = "free") +
   theme_bw() +
   theme(strip.text.y = element_text(angle = 0))

p

The solution is based on the example on page 124 in Wickham's ggplot2 book.

Solution 2:

Use scales = "free" as in:

p <- ggplot(test, aes(x=characteristic, y=es, ymin=ci_low, ymax=ci_upp)) + geom_pointrange() +
  coord_flip() + geom_hline(aes(x=0), lty=2) + 
  facet_wrap(~ set, ncol = 1, scales="free") +
  theme_bw() + 
  opts(strip.text.x = theme_text())

p

Which produces:

EDIT: I actually think I like the drop = TRUE argument better for this solution as in:

p <- ggplot(test, aes(x=characteristic, y=es, ymin=ci_low, ymax=ci_upp)) + 
  geom_pointrange() +
  coord_flip() + geom_hline(aes(x=0), lty=2) + 
  facet_wrap(~ set, ncol = 1,  drop=TRUE) +
  theme_bw() + 
  opts(strip.text.x = theme_text())

p