Clip only 1 axis using coord_cartesian

Question

I am trying to make a plot where the x-axis values are clipped (cut-off beyond the axis limits) but the y-axis values are not.

Questions:

1. Is this possible using ggplot?
2. How to code this?

Code:

Z <- data.frame(X = c(1,2,3,4,5,6,7,8), 
                Y = c(0,3,2,5,5.2,2,4,3))

# plotting with not filter (just toggle clip = on/off)
ggplot(Z, aes(x=X, y=Y)) + 
   geom_line() + 
   coord_cartesian(xlim = c(0, 7), 
                   ylim = c(0, 5), 
                   clip = 'off') + 
   theme_bw() +
   scale_y_continuous(expand = c(0,0)) +
   theme(plot.margin = unit(c(0.5, 0.5, 0.5, 0.5), 'cm'))

# plotting filter
ggplot(Z[Z$X < 7, ], aes(x=X, y=Y)) + 
   geom_line() + 
   coord_cartesian(xlim = c(0, 7), ylim = c(0, 5), clip = 'off') + 
   theme_bw() +
   scale_y_continuous(expand = c(0,0)) +
   theme(plot.margin = unit(c(0.5, 0.5, 0.5, 0.5), 'cm'))

Output (clip off/on/filtering the X values):

I want the x-axis to be stopped at x=6, but the y-values to be able to go outside the plots borders.

If I filter, then the x-axis doesn't go far enough, despite higher x-values existing. This is also the reason I have to use coord_cartesian in the first place as scale_x_continuous and xlim both disregard all future x-axis times when plotting, so stop lines earlier than they should, as descrived here

Attempt:

I can sort of get the effect I want by slightly increasing the upper y-axis limit (in this example to 5.2), but then the limit goes beyond what I want and looks odd when I plot this using my actual data.

Output of actual plot of data with the y-limit increased:

When plotting my actual data this extra line above the 100 mark doesn't make sense, because it should never go above 100%, which is why I want to avoid this workaround. (data/plot code not added for that plot for security reasons)

Answer 1

In this case, you can take advantage of the different behaviour of coord_cartesian(xlim = ...) and scale_x_continuous(limits = ...). The coords function will do a soft clipping, i.e. they draw the data and then clip it. The scales function will hard clip it, removing out of bounds values before drawing. You can be a bit more precise on the clipping behaviour of the scales functions, by setting an out-of-bounds (oob) argument to that scale. By default it will censor, i.e. remove data, but you can use the squish() function from the scales function to just have it render the points at the most extreme value.

ggplot(Z, aes(x=X, y=Y)) + 
  geom_line() + 
  coord_cartesian(ylim = c(0, 5), 
                  clip = 'off') + 
  theme_bw() +
  scale_y_continuous(expand = c(0,0)) +
  scale_x_continuous(limits = c(0, 7), expand = c(0,0), oob = scales::squish) +
  theme(plot.margin = unit(c(0.5, 0.5, 0.5, 0.5), 'cm'))

Answer 2

I might be misunderstanding your question, but can't you just filter your data?

library(tidyverse)

df <- iris # load data

ggplot(filter(df, Sepal.Width <= 4), aes(x = Sepal.Width, y = Sepal.Length)) + # plot data
    geom_point() +
    coord_cartesian(xlim = c(2, 4),    # set the x limits
                    ylim = c(4, 7.5),  # set the y limits
                    clip = 'off') +    # toggle on/off 
    theme_bw()                         # make graph a little nicer