Add secondary X axis labels to ggplot with one X axis
Solution 1
Something like this, perhaps. Note the setting of expand
for both axes to deal with proper spacing, and positions of the category names.
The labels in your figure aren't really off-centre, they are in the centre of their category boundary. It's just that by default the axes are expanded a bit further.
If you want to get more fancy, you can draw outside of the plotting area too, but it requires a bit more fiddeling. This question should get you started.
ggplot(mpg, aes(class))+
geom_bar()+
geom_text(data = data.frame(br = breaks.minor), aes(y = br, label = br, x = 7.75),
size = 4, col = 'grey30') +
coord_flip()+
scale_y_continuous(limit = lims, minor_breaks = breaks.minor,
breaks = breaks.major, labels = labels.minor,
expand = c(0, 0)) +
scale_x_discrete(expand = c(0.05, 0)) +
theme(panel.grid.major.x = element_blank()) +
theme(panel.grid.major.y = element_blank()) +
theme(axis.ticks.x=element_blank()) +
theme(axis.title= element_blank())
Solution 2
I think this does what you're looking for:
library(ggplot2)
library(grid)
library(gtable)
library(gridExtra)
breaks.major <- c(0, 15, 37.5, 52.5, 67.5, 82.5, 95, 100)
breaks.minor <- c(30, 45, 60, 75, 90)
labels.minor <- c("", "Extremely\nDissatisfied", "Dissatisfied", "Uncertain",
"Satisfied", "Very\nSatisfied", "Extremely\nSatisfied", "")
lims <- c(0, 100)
# build the main plot with the text axis
gg1 <- ggplot(mpg, aes(class))
gg1 <- gg1 + geom_bar()
gg1 <- gg1 + scale_y_continuous(expand=c(0,0), limit=lims,
minor_breaks=breaks.minor,
breaks=breaks.major,
labels=labels.minor)
gg1 <- gg1 + coord_flip()
gg1 <- gg1 + theme(panel.grid.major.x=element_blank())
gg1 <- gg1 + theme(panel.grid.major.y=element_blank())
gg1 <- gg1 + theme(axis.ticks.x=element_blank())
gg1 <- gg1 + theme(axis.title=element_blank())
# let ggplot2 do the work of building the second axis
gg2 <- ggplot(mpg, aes(class))
gg2 <- gg2 + scale_y_continuous(expand=c(0,0), limit=lims,
breaks=c(0, breaks.minor, 100))
gg2 <- gg2 + coord_flip()
gg2 <- gg2 + theme(axis.ticks.x=element_blank())
gg2 <- gg2 + theme(axis.text.x=element_text(hjust=c(0, 0.5, 0.5, 0.5, 0.5, 0.5, 1)))
gt1 <- ggplot_gtable(ggplot_build(gg1))
gt2 <- ggplot_gtable(ggplot_build(gg2))
axis2 <- grid.arrange(gt2$grobs[[5]])
gt <- gtable_add_rows(gt1, unit(0.1, "null"), 4)
grid.arrange(gtable_add_grob(gt, axis2, t=5, l=4, b=5, r=4))
Solution 3
I wrote all labels as major-labels.
# OP's
breaks.major <- c(0,15,37.5,52.5,67.5,82.5,95,100) #defines the midpoints of the categories (label locations)
breaks.minor <- c(30,45,60,75,90) #defines the edges of the categories (second label set I need)
labels.minor <- c("","Extremely \nDissatisfied","Dissatisfied","Uncertain","Satisfied","Very \nSatisfied","Extremely \nSatisfied","")
lims =c(0,100)
breaks.major2 <- c(0,15,37.5,52.5,67.5,82.5,95)
breaks.minor2 <- c(30,45,60,75,90,100) # put 100 into minor from major
breaks.comb <- sort(c(breaks.major2, breaks.minor2 - 1.0E-6)) # avoid the just same value as minor
label.comb <- c(0, "\nExtremely \nDissatisfied", 30, "\nDissatisfied", 45, "\nUncertain", 60,
"\nSatisfied", 75, "\nVery \nSatisfied", 90, "\nExtremely \nSatisfied", 100)
library(ggplot2)
g <- ggplot(mpg, aes(class))+
geom_bar()+
coord_flip()+
scale_y_continuous(limit = lims, minor_breaks = breaks.minor2, breaks = breaks.comb,
labels = label.comb, expand = c(0,0)) +
theme(panel.grid.major.x = element_blank()) +
theme(panel.grid.major.y = element_blank()) +
theme(axis.ticks.x=element_blank()) +
theme(axis.title= element_blank()) +
theme(plot.margin = unit(c(0.5, 0.5, 0.5, 0.5), "lines"))
Alex
PhD student studying visual sciences in Melbourne, Australia
Updated on June 17, 2022Comments
-
Alex almost 2 years
**Edit, there are two great solutions here, one is marked as the answer, but @hrbrmstr provides a great solution combining two ggplots which works well for this simple plot.*
Here's the code
breaks.major <- c(0,15,37.5,52.5,67.5,82.5,95,100) #defines the midpoints of the categories (label locations) breaks.minor <- c(30,45,60,75,90) #defines the edges of the categories (second label set I need) labels.minor <- c("","Extremely \nDissatisfied","Dissatisfied","Uncertain","Satisfied","Very \nSatisfied","Extremely \nSatisfied","") lims =c(0,100) g <- ggplot(mpg, aes(class))+ geom_bar()+ coord_flip()+ scale_y_continuous(limit = lims, minor_breaks = breaks.minor, breaks = breaks.major, labels = labels.minor) + theme(panel.grid.major.x = element_blank()) + theme(panel.grid.major.y = element_blank()) + theme(axis.ticks.x=element_blank()) + theme(axis.title= element_blank())
It produces this plot:
I need to have two sets of X-axis labels, one showing the category names (i.e. the "satisfied" etc. that are already there via
labels.minor
), and one showing the values at thebreaks.minor
locations (corresponding to the category limits, i.e. the vertical panel grid lines). I need the currentlabels.minor
labels to be below the required additional labels.I currently do this with line breaks so that the numbers and categories are all in one long string, but the spacing gets funny with plot resizes.I could do this with text boxes (I assume), is there a way within ggplot?
Extra points if you get my current labels in the centre of their sections (e.g. "Extremely Satisfied" is off-centre)
This is my desired output (pardon my 'mspaint')
-
Alex over 7 yearsAwesome @Axeman, thanks for the effort and the link I'll continue on. I do, unfortunately, need the numbers between the plot and the 'satisfied' etc however (see updated figure in op). I probably wasn't clear on that, though your suggestion might also have been an alternative approach given limitations in ggplot to do this natively.
-
Axeman over 7 yearsJust replace
x
ingeom_text
with something smaller to move the numbers down to get them close. -
Axeman over 7 yearsYou could also just add all of those as
breaks
, and use line breaks (\n
) to move some labels up or down. -
Alex over 7 yearsof course RE changing
x
, silly oversight. And as to usingbreaks
, that was how this code was originally, as one long character string with spaces and breaks to line everything up. Sadly it needs to be adjusted as the plot changes size occasionally. -
Alex over 7 yearsExcellent solution that works very well with this example code. Sadly my actual plot code is much longer, and has conditional error bars and other complicated formatting. Effectively doubling the code for the plotting routine would blow things out. Can I mark two answers as correct?Because this technically is.
-
hrbrmstr over 7 yearsI'm fairly certain this can be made a bit more minimalistic. I'll give it a go later this eve.
-
Alex over 7 yearsHey mate, i'm getting weird issues when using this when I split my bars using
fill
in theggplot
call. Unless the length ofbreaks.minor
is the same as the length of the labels for the y axis (aftercoord_flip
) I get an errorError: Aesthetics must be either length 1 or the same as the data (12): x, y, label, fill
. So bizarre, how can I have thegeom_text
act independently of the data in the plot? It feels like this should be easy!!! -
Alex over 7 yearsHey @hrbrmstr did you figure out how to condense this?