How to expand axis asymmetrically with ggplot2 without setting limits manually?

Question

I want the bottom axis for a barplot to be zero, but top expanded by a small amount (according to some scale). However, I am plotting the output of stat_summary so I do not have the maximum value available to me, and I am using facets with scale="free_y" so I cannot manually set them. Otherwise I would use baptiste's trick of adding geom_blank. I know I could read off the values and create a data frame with upper bounds manually, but can this be set without having to hard-code the limits?

The other option is to compute the means outside of the function call and just plot that without calling stat_summary (this way I have the upper bounds directly for each facet), but is there any way around this?

Example:

ggplot(mtcars)+
  stat_summary(aes(cyl,mpg),
               fun.y="mean",
               geom="bar")+
  scale_y_continuous(expand=c(0,0))+
  facet_grid(carb~.,"free_y")

Answer 1

You can "extend" ggplot by creating a scale with a custom class and implementing the internal S3 method scale_dimension like so:

library("scales")
scale_dimension.custom_expand <- function(scale, expand = ggplot2:::scale_expand(scale)) {
  expand_range(ggplot2:::scale_limits(scale), expand[[1]], expand[[2]])
}

scale_y_continuous <- function(...) {
  s <- ggplot2::scale_y_continuous(...)
  class(s) <- c('custom_expand', class(s))
  s
}

This is also an example of overriding the default scale. Now, to get the desired result you specify expand like so

qplot(1:10, 1:10) + scale_y_continuous(expand=list(c(0,0.1), c(0,0)))

where the first vector in the list is the multiplicative factor, the second is the additive part and the first (second) element of each vector corresponds to lower (upper) boundary.