arranging order of stacked geom_bar in ggplot2

Question

I am drawing some pie charts using facet_wrap and coord_polar, let say I have factors with sequence A,B,C.D, for the x-axis, and by using the facet_wrap (for 3 charts, let say) and coord_polar. I will have 3 pies (pie X,Y,Z) with arranged as A,B,C.D in clockwise direction.

Problem arise when I want to re-arrange the level of the factor individually for each facet, e.g.:

   A  B  C  D
X 10 20 30 40
Y 20 30 40 10
Z 30 40 10 20

I want:

pie X to be arranged as : A,B,C.D clockwise

pie Y to be arranged as : D,A,B.C clockwise

pie Z to be arranged as : C,D,A.B clockwise

Can I do that in ggplot2? Thanks.

I think you'd be better off doing the 3 plots separately and arranging the plots on a single device using some grid functions to push plots onto the viewport. ggplot expects the data to come as coherent chunk and that precludes changing the orders of factors on the fly, internally, for different panels. If you are interested in the grid viewport stuff, I can supply an answer along those lines? — Gavin Simpson, Apr 19 '11 at 10:58
possible duplicate of [How to order bars in faceted ggplot2 bar chart](http://stackoverflow.com/questions/5409776/how-to-order-bars-in-faceted-ggplot2-bar-chart) — Richie Cotton, Apr 19 '11 at 11:08

score 7 · Accepted Answer · edited Apr 19 '11 at 12:13

The grid/viewport approach would work something like this. First we arrange some data to match what you show. We rotate the levels of the factor for each of X, Y and Z as you specify in the question

## Your data
dat <- data.frame(X = factor(rep(LETTERS[1:4], times = c(10,20,30,40)), 
                             levels = LETTERS[1:4]),
                  Y = factor(rep(LETTERS[1:4], times = c(20,30,40,10)),
                             levels = LETTERS[c(4,1:3)]),
                  Z = factor(rep(LETTERS[1:4], times = c(30,40,10,20)),
                             levels = LETTERS[c(3:4,1:2)]))

Next we produce the individual pie plots, rotating the same scale_fill_manual() colours in the same order we rotated the levels in the data, dat (with apologies for the nasty colours - how does one get the first n colours that ggplot would use normally?):

p1 <- ggplot(dat, aes(x = factor(1), fill = factor(X))) + 
          geom_bar(width = 1) + coord_polar(theta = "y") +
          scale_fill_manual(value = c("red","green","yellow","blue"))
p2 <- ggplot(dat, aes(x = factor(1), fill = factor(Y))) + 
          geom_bar(width = 1) + coord_polar(theta = "y") +
          scale_fill_manual(value = c("red","green","yellow","blue")[c(4,1:3)])
p3 <- ggplot(dat, aes(x = factor(1), fill = factor(Z))) + 
          geom_bar(width = 1) + coord_polar(theta = "y") +
          scale_fill_manual(value = c("red","green","yellow","blue")[c(3:4,1:2)])

Now we need a new plotting page and to push onto it a 2x2 layout:

grid.newpage()
pushViewport(viewport(layout = grid.layout(2,2)))

A custom function allows us to simplify setting the row/col for each subsequent plot:

vplayout <- function(x, y) {
    viewport(layout.pos.row = x, layout.pos.col = y)
}

Next we simply plot each save ggplot object on the correct viewport, as specified by argument vp:

print(p1, vp = vplayout(1,1))
print(p2, vp = vplayout(1,2))
print(p3, vp = vplayout(2,1))

The above grid code was taken from page 94 in Hadley's excellent book on ggplot.

This produces:

pie charts and viewport layout

arranging order of stacked geom_bar in ggplot2

1 Answers1