NA handling: methods for missing data

[Nicholas Lewin-Koh]

Hi Hadley,
After searching through the documentation, we realized that the default
handling of NA's
in ggplot2 is "na.omit". It would be nice if there were some ways to
handle them differently
in ggplot. Something like
p <- ggplot(mtcars, aes(mpg, wt, colour = cyl)) + geom_point()
p + NA_geom_rug(aes(mpg,wt))

Or something like that, where NAs could be handled explicitly. This is
probably possible by just doing separate layers for each axis, but
explicit methods
would be nice. Perhaps this is time for me to roll up my sleeves and
submit
something to ggplot extra.

Nicholas

[hadley wickham]

Hi Nicholas,

It varies a little bit - for categorical scales, missings will always
be included as one of the categories. There's no equivalent for
continuous scales. I've thought about a scale that is a combination
of continuous and categorical, but it would be a lot of work.

Hadley

--
http://had.co.nz/