Plotting labels instead of markers

[urisohn]

Been using beeswarm for a while (e.g., in our blog Data Colada). Now using it for a new pkg that creates plots for reporting results from psychology experiments, and one of the plots we wanted to offer was the beeswarm plot, but using text-labels instead of markers.
Like. Imagine with data at the State level in the US and every 'marker' is the two letter abbreviation for the state (AZ, AK, etc.)

You can see an example inm the uploaded plot. I ended up creating a beeswarm, without plotting, using the exprted values, and then using text() to populate it, but maybe it could be built into the package as an option. Just a suggestion.

[aroneklund]

Hmm. That's not a bad idea.

I guess you could create a simple version with something like this (which I think is what you are describing):

out <- beeswarm(state.area, log = TRUE, spacing = 4, pch = NA)
text(out$x, out$y, labels = state.abb)

Here I set spacing = 4 based on trial and error. I think this works pretty well with state abbreviations, since they are all around the same size.

Are you proposing a more automatic approach, such as taking the size of each individual label into account when doing the layout? (maybe similar to #5 but with text instead of points?)

[urisohn]

Yes, that's basically what i did, but i think some automatic spacing would be needed to prevent overstepping too much
i find it hard to navigate the mapping between text and absolute dimensions so could be tricky

The hacky way I did was to compute the average length of the labels and convert that to the horizontal spacing.
LIke this

 stimulus.length = mean(nchar(ms$stimulus))
  dot.spacing = stimulus.length/3 + 2

where ms$stimulus contains the labels you see, and then i make spacing=dot.spacing inside the beeswarm call