Has anyone done this before, and do you have any reflections on font
size, colour choices, etc. that you could offer? Obviously Tufte's
guidance applies and less is almost always better, and while Hadley has
done a fine job with the defaults, this is a very different graphical
display than paper with high resolution printing. So, experience with
this format of presentation would be most welcome.
Most of the data will be bar charts, box plots, a couple of X/Y graphs,
and some use of geom_crossbar to depict graphically beta estimates and
standard errors from logistic regression models (which, by the way, is a
lovely way to get people to focus on the overall shape of the model
rather than on whether the p value is 0.04 or 0.06).
Suggestions welcome.
Thanks
Coming back to the question, you might want to try,
- define your own custom theme (perhaps: horizontal y-axis labels,
large sans serif font, white background, no box, etc.). It would be
great if you could share such a theme with us afterwards.
- the dichromat colour scales from the dichromat package (an
experimental interface à la scale_colour_brewer() for ggplot2 is in
the package ggextra)
If you were to make the whole presentation by yourself, I'd suggest
you try the tikzDevice package and Sweave + Beamer, but it looks as
though you'll be using powerpoint.
All the best,
baptiste