gghighlight 0.3.2

n()

Since gghighlight uses dplyr inside, you can now use dplyr’s expression, n(). This is useful to highlight based on the size of the group.

Suppose we have this data:

library(gghighlight)

Loading required package: ggplot2

library(dplyr, warn.conflicts = FALSE)

set.seed(1098)
centers <- tibble(
  id = sample(letters, 11),
  x = c(-1, -2, -3,   0,  1,  4,  2,  5,  7,  1, -1),
  y = c( 4, -3, -7, -10, -8, -3,  9,  5, -1,  1, -1),
  n = c(50, 50, 100, 50, 50, 120, 40, 10, 20, 5, 8)
)

d <- centers %>% 
  rowwise() %>% 
  summarise(id = id, x = x + rnorm(n, sd = 3), y = y + rnorm(n, sd = 3))

p <- ggplot(d, aes(x, y, colour = id)) + geom_point()
p

By using n(), we can focus on the large groups.

p +
  gghighlight(n() >= 100, use_direct_label = FALSE)

Or small groups.

p +
  gghighlight(n() < 10, use_direct_label = FALSE)

You can also use n() as a non-logical predicate, whose values are used for sorting data and the top max_highlight of rows/groups are highlighted.

# Same result as above
p +
  gghighlight(-n(), max_highlight = 2, use_direct_label = FALSE)

To unhighlight or not to unhighlight…

By default, unhighlighted data are grayed out. unhighlighted_params is the option to override this. Now, you can even choose not to unhighlight at all by specifying explicit NULL to colour or fill!

p +
  gghighlight(n() < 10, use_direct_label = FALSE,
              unhighlighted_params = list(colour = NULL))

Hmm…, but this is the very same plot as the original one. How can this be useful? Well, remember we still can tweak other parameters like alpha.

p +
  gghighlight(n() < 10, use_direct_label = FALSE,
              unhighlighted_params = list(colour = NULL, alpha = 0.2))

This plot doesn’t look very nice in that the colors are a bit difficult to distinguish. This is mainly because I didn’t come up with some nice data, but it’s generally a tough job to tweak colors by alpha properly, so I don’t recommend this much. But, hope you can find some good use case for this!