Question

I’m struggling a bit with ggplot2 and hoping to learn by example a bit more.

I have a lot of data that looks like like the thing this generates:

data.frame(version=c('v1', 'v1', 'v1', 'v1', 'v2', 'v2', 'v2', 'v2'),
           platform=c('linux', 'linux', 'mac', 'mac',
                      'linux', 'linux', 'mac', 'mac'),
           type=c('a', 'b', 'a', 'b', 'a', 'b', 'a', 'b'),
           count=floor(runif(8, 0, 10000)))

I got regular barplot to draw me a stacked bar chart of type for a given OS (by slicing it out with cast, but I’ve not quite got what I want with ggplot2 yet.

I can manage to plot a single platform by doing something like this (assuming the above is saved as sample):

qplot(version, a, data=cast(sample[sample$platform=='linux',],
                            version ~ type, value="count"),
      geom='bar')

Ideally, I’d like that stacked by type (explicitly a in this example — there are only two types) and then have one per platform appear side-by-side on the same chart grouped by version.

That is, for every version, I’d like three bars (one for each platform) with two stacks each (by type).

Answer 1

Here’s one option:

dat <- data.frame(version=c('v1', 'v1', 'v1', 'v1', 'v2', 'v2', 'v2', 'v2'),
           platform=c('linux', 'linux', 'mac', 'mac',
                      'linux', 'linux', 'mac', 'mac'),
           type=c('a', 'b', 'a', 'b', 'a', 'b', 'a', 'b'),
           count=floor(runif(8, 0, 10000)))


ggplot(data = dat, aes(x = platform, y = count)) + 
    facet_wrap(~version) + 
    geom_bar(aes(fill = type))

which produces something like this:

Your example data only had two platforms, so maybe that was just a typo since you referenced three.