facet_grid can't deal with more than two facetting variables per margin #378
Description
Example:
d <- expand.grid(a=1:2, b=3:4, c=c(5:6, 5:6), d=7:8)
d$value <- rnorm(nrow(d))
d$x <- 1
ggplot(d, aes(x=x, y=value)) + geom_point() + facet_grid(a + b + c ~ d)
## throws an error
## stupid workaround:
d$bc <- with(d, factor(b):factor(c))
ggplot(d, aes(x=x, y=value)) + geom_point() + facet_grid(a + bc ~ d)EDIT:
BrianDiggs has a good sugestion below, mine is overcomplicated.
There is also the simple workaround of simply doing:
facet_grid(list(c("a","b","c"), "d"))
Suggested solution:
the problem is the way that facet_grid defines the rows and cols simply as
rows <- as.quoted(facets[[2]])
which breaks down for a formula with more than two terms.
Instead, I would suggest using a real formula parser to actually find all the terms, i.e.
use
getVars <- function(frm){
if(length(as.quoted(frm))==1) {
if(as.character(frm)!="+"){
return(as.quoted(frm))
} else return(NULL)
} else {
return(c(getVars(frm[[1]]), getVars(frm[[2]]), as.quoted(frm[[3]])))
}
}
and replace lines
rows <- as.quoted(facets[[2]])
cols <- as.quoted(facets[[3]])
in facet_grid with
rows <- getVars(facets[[2]])
cols <- getVars(facets[[3]])