Simple text labels
Markdown-formatted text labels can be placed into a plot with
geom_richtext(). This geom is mostly a drop-in replacement
for geom_label() (or geom_text()), with added
capabilities.
As a first example, we will annotate a plot of linear regressions
with their r2 values. We will use the
iris dataset for this demonstration. In our first
iteration, we will not yet use any ggtext features, and instead plot the
text with geom_text().
library(ggplot2)
library(dplyr)
library(glue)
iris_cor <- iris %>%
group_by(Species) %>%
summarize(r_square = cor(Sepal.Length, Sepal.Width)^2) %>%
mutate(
# location of each text label in data coordinates
Sepal.Length = 8, Sepal.Width = 4.5,
# text label containing r^2 value
label = glue("r^2 = {round(r_square, 2)}")
)
iris_cor
#> # A tibble: 3 × 5
#> Species r_square Sepal.Length Sepal.Width label
#> <fct> <dbl> <dbl> <dbl> <glue>
#> 1 setosa 0.551 8 4.5 r^2 = 0.55
#> 2 versicolor 0.277 8 4.5 r^2 = 0.28
#> 3 virginica 0.209 8 4.5 r^2 = 0.21
iris_facets <- ggplot(iris, aes(Sepal.Length, Sepal.Width)) +
geom_point() +
geom_smooth(method = "lm", formula = y ~ x) +
facet_wrap(~Species) +
theme_bw()
iris_facets +
geom_text(
data = iris_cor,
aes(label = label),
hjust = 1, vjust = 1
)
This code works, but the result is not fully satisfying. First,
because r is a mathematical variable, it should be typeset in
italics. Second, it would be nicer to have a superscript 2 instead of
^2. We can achieve both results by creating a markdown label and
plotting it with geom_richtext().
library(ggtext)
iris_cor_md <- iris_cor %>%
mutate(
# markdown version of text label
label = glue("*r*<sup>2</sup> = {round(r_square, 2)}")
)
iris_cor_md
#> # A tibble: 3 × 5
#> Species r_square Sepal.Length Sepal.Width label
#> <fct> <dbl> <dbl> <dbl> <glue>
#> 1 setosa 0.551 8 4.5 *r*<sup>2</sup> = 0.55
#> 2 versicolor 0.277 8 4.5 *r*<sup>2</sup> = 0.28
#> 3 virginica 0.209 8 4.5 *r*<sup>2</sup> = 0.21
iris_facets +
geom_richtext(
data = iris_cor_md,
aes(label = label),
hjust = 1, vjust = 1
)
By default, geom_richtext() puts a box around the text
it draws. We can suppress the box by setting the fill and outline colors
to transparent (fill = NA, label.colour = NA).
iris_facets +
geom_richtext(
data = iris_cor_md,
aes(label = label),
hjust = 1, vjust = 1,
# remove label background and outline
fill = NA, label.color = NA,
# remove label padding, since we have removed the label outline
label.padding = grid::unit(rep(0, 4), "pt")
)
We can separately choose the colors of label outline, label fill, and label text, and we can assign them via aesthetic mapping as well as by direct specification, as is usual in ggplot2.
iris_facets +
aes(colour = Species) +
geom_richtext(
data = iris_cor_md,
aes(
label = label,
fill = after_scale(alpha(colour, .2))
),
text.colour = "black",
hjust = 1, vjust = 1
) +
theme(legend.position = "none")
Rotated labels are also possible, though in most cases it is not recommended to use them.
iris_facets +
aes(colour = Species) +
geom_richtext(
data = iris_cor_md,
aes(
x = 7.5,
label = label,
fill = after_scale(alpha(colour, .2))
),
text.colour = "black",
hjust = 1, vjust = 1,
angle = 30
) +
theme(legend.position = "none")