I am trying to produce a plot with a sort of "hybrid" layout between facet_grid and facet_wrap:
example:
library(reshape2)
library(ggplot2)
data(diamonds)
# sample and reshape diamond dataset
diamonds_l <- melt(diamonds[sample(1:nrow(diamonds), 200),
c("cut", "depth", "table", "price", "x")],
id.vars = c("cut","x"))
this is the plot arrangement that I want (quality in columns and depth, table, price as rows)
ggplot( diamonds_l, aes( x = value, y = x, colour= cut))+
geom_point( ) +
facet_wrap( variable ~ cut, scales = "free_x", nrow=3)
however, I would prefer the facet_grid design (only one header per column / row) but scales = "free_x" doesn't work in this layout
ggplot( diamonds_l, aes( x = value, y = x, colour= cut))+
geom_point( ) +
facet_grid(variable ~ cut, scales = "free_x")
it works here but thats not the arrangement that I want (quality as rows)
ggplot( diamonds_l, aes( x = value, y = x, colour= cut))+
geom_point( ) +
facet_grid(cut ~ variable, scales = "free_x")
I understand why it won't work but I was wondering if there was a work-around?
thanks!
Fabian
After some digging, I managed to pull the trick. Pieces of code borrowed from here (@baptiste, @Roland) and here (@Sandy Muspratt). Looks scary, but I'm basically removing/renaming text strips from the first plot of yours via low-level manipulations.
p <- ggplot( diamonds_l, aes( x = value, y = x, colour= cut))+
geom_point( ) +
facet_wrap( variable ~ cut, scales = "free_x", nrow = 3)
library(gridExtra)
gt <- ggplotGrob(p)
panels <- grep("panel", gt$layout$name)
top <- unique(gt$layout$t[panels])
gt <- gt[-(top[-1]-1), ]
gg <- gt$grobs
strips <- grep("strip_t", names(gg))
labels <- levels(diamonds_l$cut)
for(ii in seq_along(labels)) {
modgrob <- getGrob(gg[[strips[ii]]], "strip.text",
grep=TRUE, global=TRUE)
gg[[strips[ii]]]$children[[modgrob$name]] <- editGrob(modgrob,label=labels[ii])
}
gt$grobs <- gg
grid.newpage()
grid.draw(gt)