I have some data that I've divided into enough groupings that standard boxplots look very crowded. Tufte has his own boxplots in which you basically drop all or half of box, like this:
Some sample data:
cw <- transform(ChickWeight,
Time = cut(ChickWeight$Time,4)
)
cw$Chick <- as.factor( sample(LETTERS[seq(3)], nrow(cw), replace=TRUE) )
levels(cw$Diet) <- c("Low Fat","Hi Fat","Low Prot.","Hi Prot.")
I want a boxplot of weight for every Diet * Time * Chick grouping.
I had this problem come up years ago, and kludged together a solution using grid graphics, which I'll post in a bit. But in solving this new (and similar) problem I'm wondering if there's a stock way to do them rather than fixing my kludged together example.
As an aside, these seem to be amongst the less-beloved of Tufte's creations, but I really like them for densely displaying patterns of distributions across a large number groupings, and I'd use them more if there was a good function for them in ggplot2 or lattice.