Apply coord_flip() to single layer

Question

I would like to have a boxplot showing the same distribution underneath my histogram. The code below almost works, but coord_flip() is being applied to all layers, instead of just the geom_boxplot layer.

plot1<-ggplot(newdatahistogram, aes_string(x=newdatahistogram[RawLocation])) + 
          xlab(GGVar) + ylab("Proportion of Instances") + 
          geom_histogram(aes(y=..density..), binwidth=1, colour="black", fill="white",origin=-0.5) + 
          scale_x_continuous(limits=c(-3,6), breaks=seq(0,5,by=1), expand=c(.01,0)) +
          geom_boxplot(aes_string(x=-1, y=newdatahistogram[RawLocation])) + coord_flip()

How can I apply coord_flip() to a single layer?

Thank you!

Please use the examples here to ask a good reproducible question: http://stackoverflow.com/questions/5963269/how-to-make-a-great-r-reproducible-example — Drew Steen, Sep 08 '13 at 00:14
Have a look at https://groups.google.com/forum/#!topic/ggplot2/uFKOFcZSOIU — mnel, Sep 09 '13 at 03:45

score 2 · Accepted Answer · answered Sep 09 '13 at 03:33

I got it to work with a bit of a hack;

plot1 <- ggplot(newdatahistogram, aes_string(x=newdatahistogram[RawLocation], fill=(newdatahistogram[,"PQ"]))) + 
  xlab(GGVar) + ylab("Proportion of Observation") + 
  geom_histogram(aes(y=..density..), binwidth=1, colour="black", origin=-0.5) +
  scale_x_continuous(limits=c(-1,6), breaks=seq(0,5,by=1), expand=c(.01,0)) +
  scale_y_continuous(limits=c(-.2,1), breaks=seq(0,1,by=.2))
  theme(plot.margin = unit(c(0,0,0,0), "cm"))      

plot_box <- ggplot(newdatahistogram) +
  geom_boxplot(aes_string(x=1, y=newdatahistogram[RawLocation])) + 
  scale_y_continuous(breaks=(0:5), labels=NULL, limits=c(-1,6), expand=c(.0,-.03)) + 
  scale_x_continuous(breaks=NULL) + xlab(NULL) + ylab(NULL) +
  coord_flip() + theme_bw() +  
  theme(plot.margin = unit(c(0,0,.0,0), "cm"),
        line=element_blank(),text=element_blank(),
        axis.line = element_blank(),title=element_blank(), panel.border=theme_blank())

PB = ggplotGrob(plot_box)
plot1 <- plot1 + annotation_custom(grob=PB, xmin=-1.01, xmax=5.95, ymin=-.3,ymax=0)

This saves the rotated boxplot as a grob object and inserts it into the plot under the histogram.

I needed to play with the expansion element a bit to get the scales to line up, but it works!

Seriously though, I think ggplot should have a horizontal boxplot available without cord_flip()... I tried to edit the boxplot code, but it was way too difficult for me!

Tried to post image, but not enough reputation

score 1 · Answer 2 · edited May 23 '17 at 12:13

1

You can't: coord_flip always acts on all layers. However, you do have two alternatives:

The solution here shows how to use grid.arrange() to add a marginal histogram. (The comments in the question also link to a nice base-R way to do the same thing)
You could indicate density using a rug plot on of the four sides of the plot with plot1 + geom_rug(sides='r')

ggplot(mpg, aes(x=class, y=cty)) + geom_boxplot() + geom_rug(sides="r")

enter image description here

edited May 23 '17 at 12:13

Community

1
1

answered Sep 08 '13 at 00:05

Drew Steen

16,045
12
62
90

Hi Drew, this is the 2nd time you've answered one of my questions, and I want you to know I really appreciate it! Thank you! – JayCo Sep 09 '13 at 03:38

Apply coord_flip() to single layer

2 Answers2