How can I make a discontinuous axis in R with ggplot2?

Question

I have a dataframe (dat) with two columns 1) Month and 2) Value. I would like to highlight that the x-axis is not continuous in my boxplot by interrupting the x-axis with two angled lines on the x-axis that are empty between the angled lines.

Example Data and Boxplot

library(ggplot2)
set.seed(321)
dat <- data.frame(matrix(ncol = 2, nrow = 18))
x <- c("Month", "Value")
colnames(dat) <- x
dat$Month <- rep(c(1,2,3,10,11,12),3)
dat$Value <- rnorm(18,20,2)

ggplot(data = dat, aes(x = factor(Month), y = Value)) +
  geom_boxplot() +
  labs(x = "Month") +
  theme_bw() +
  theme(panel.grid = element_blank(),
        text = element_text(size = 16),
        axis.text.x = element_text(size = 14, color = "black"),
        axis.text.y = element_text(size = 14, color = "black"))

The ideal figure would look something like below. How can I make this discontinuous axis in ggplot?

Answer 1

You could make use of the extended axis guides in the ggh4x package. Alas, you won't easily be able to create the "separators" without a hack similar to the one suggested by user Zhiqiang Wang

guide_axis_truncated accepts vectors to define lower and upper trunks. This also works for units, by the way, then you have to pass the vector inside the unit function (e.g., trunc_lower = unit(c(0,.45), "npc") !

library(ggplot2)
library(ggh4x)

set.seed(321)
dat <- data.frame(matrix(ncol = 2, nrow = 18))
x <- c("Month", "Value")
colnames(dat) <- x
dat$Month <- rep(c(1,2,3,10,11,12),3)
dat$Value <- rnorm(18,20,2)

# this is to make it slightly more programmatic
x1end <- 3.45
x2start <- 3.55

p <-
ggplot(data = dat, aes(x = factor(Month), y = Value)) +
  geom_boxplot() +
  labs(x = "Month") +
  theme_classic() +
  theme(axis.line = element_line(colour = "black"))

p +
  guides(x = guide_axis_truncated(
    trunc_lower = c(-Inf, x2start),
    trunc_upper = c(x1end, Inf)
  ))

^{Created on 2021-11-01 by the reprex package (v2.0.1)}

The below is taking user Zhiqiang Wang's hack a step further. You will see I am using simple trigonometry to calculate the segment coordinates. in order to make the angle actually look as it is defined in the function, you would need to set coord_equal.

# a simple function to help make the segments
add_separators <- function(x, y = 0, angle = 45, length = .1){
  add_y <-  length * sin(angle * pi/180)
  add_x <- length * cos(angle * pi/180)
  ## making the list for your segments
  myseg <- list(x = x - add_x, xend = x + add_x, 
                y = rep(y - add_y, length(x)), yend = rep(y + add_y, length(x)))
  ## this function returns an annotate layer with your segment coordinates
  annotate("segment", 
           x = myseg$x, xend = myseg$xend,
           y = myseg$y, yend = myseg$yend) 
}

# you will need to set limits for correct positioning of your separators
# I chose 0.05 because this is the expand factor by default 
y_sep <- min(dat$Value) -0.05*(min(dat$Value))

p +
  guides(x = guide_axis_truncated(
    trunc_lower = c(-Inf, x2start),
    trunc_upper = c(x1end, Inf)
  )) +
  add_separators(x = c(x1end, x2start), y = y_sep, angle = 70) +
  # you need to set expand to 0
  scale_y_continuous(expand = c(0,0)) +
  ## to make the angle look like specified, you would need to use coord_equal()
  coord_cartesian(clip = "off", ylim = c(y_sep, NA)) 

Answer 2

I think it is possible to get what you want. It may take some work.

Here is your graph:

library(ggplot2)
set.seed(321)
dat <- data.frame(matrix(ncol = 2, nrow = 18))
x <- c("Month", "Value")
colnames(dat) <- x
dat$Month <- rep(c(1,2,3,10,11,12),3)
dat$Value <- rnorm(18,20,2)

p <- ggplot(data = dat, aes(x = factor(Month), y = Value)) +
  geom_boxplot() +
  labs(x = "Month") +
  theme_bw() +
  theme(panel.grid = element_blank(),
        text = element_text(size = 16),
        axis.text.x = element_text(size = 14, color = "black"),
        axis.text.y = element_text(size = 14, color = "black"))

Here is my effort:

p + annotate("segment", x = c(3.3, 3.5), xend = c(3.6, 3.8), y = c(14, 14), yend = c(15, 15))+
  coord_cartesian(clip = "off", ylim = c(15, 25)) 

Get something like this:

If you want to go further, it may take several tries to get it right:

p + annotate("segment", x = c(3.3, 3.5), xend = c(3.6, 3.8), y = c(14, 14), yend = c(15, 15))+
  annotate("segment", x = c(0, 3.65), xend = c(3.45, 7), y = c(14.55, 14.55), yend = c(14.55, 14.55)) +
  coord_cartesian(clip = "off", ylim = c(15, 25)) +
  theme_classic()+
  theme(axis.line.x = element_blank())
  

Just replace axis with two new lines. This is a rough idea, it may take some time to make it perfect.

Answer 3

You could use facet_wrap. If you assign the first 3 months to one group, and the other months to another, then you can produce two plots that are side by side and use a single y axis.

It's not exactly what you want, but it will show the data effectively, and highlights the fact that the x axis is not continuous.

dat$group[dat$Month %in% c("1", "2", "3")] <- 1
dat$group[dat$Month %in% c("10", "11", "12")] <- 2

ggplot(data = dat, aes(x = factor(Month), y = Value)) +
  geom_boxplot() +
  labs(x = "Month") +
  theme_bw() +
  theme(panel.grid = element_blank(),
        text = element_text(size = 16),
        axis.text.x = element_text(size = 14, color = "black"),
        axis.text.y = element_text(size = 14, color = "black")) +
  facet_wrap(~group, scales = "free_x")

^{* _{Differences in the plot are likely due to using different versions of R where the set.seed gives different result}}