Annotations by Nicolas Nova is licensed under
CC
BY-NC 2.0
introduction
highlight data symbols
highlight data with labels
text placed arbitrarily
text placed arbitrarily in multiple
panels
references
By annotation, I mean adding text or points to a graph to draw the reader’s attention to specific data.
One consistent approach is to create a data frame for the annotation and use it as the data argument in a geom.
Sometimes the coordinates of the annotation depend on the data; other times the annotation has arbitrary coordinates.
library("tidyverse")
library("graphclassmate")
library("gapminder")
library("ggrepel")Suppose we have this graph
df <- gapminder %>%
filter(year == 2007) %>%
filter(continent != "Oceania") %>%
select(-year) %>%
mutate(continent = fct_reorder(continent, lifeExp)) %>%
droplevels()
p <- ggplot(df, aes(x = gdpPercap/10000, y = lifeExp)) +
geom_point(alpha = 0.6) +
facet_wrap(vars(continent), as.table = FALSE, ncol = 1) +
labs(x = "GDP per capita ($10k ppp)", y = "", title = "Life expectancy (years)", caption = "Source: gapminder")
p
And we would like to highlight the country with highest life expectancy in each group. We extract that subset of the data.
high_four <- df %>%
group_by(continent) %>%
filter(lifeExp == max(lifeExp))
high_four
#> # A tibble: 4 x 5
#> # Groups: continent [4]
#> country continent lifeExp pop gdpPercap
#> <fct> <fct> <dbl> <int> <dbl>
#> 1 Canada Americas 80.7 33390141 36319.
#> 2 Iceland Europe 81.8 301931 36181.
#> 3 Japan Asia 82.6 127467972 31656.
#> 4 Reunion Africa 76.4 798094 7670.We use this data frame as the data argument in a new geom_point()
layer.
p1 <- p + geom_point(data = high_four,
aes(x = gdpPercap/10000, y = lifeExp),
color = "red", size = 3)
p1
To label the country, use could use geom_text(), but the printing
overlaps the data points.
p2 <- p1 + geom_text(data = high_four,
aes(x = gdpPercap/10000, y = lifeExp, label = country))
p2
Try the ggrepel package to move the labels away from the data points.
This takes a little bit of work. First, I create a data frame that has
an empty character string for all country names except the four I want.
For geom_text_repel() to avoid all the data points, we have to print
all the countries. But in this new data frame, all country names are
blank except the four I want.
Because I’m adding a new value ("") to the country factor I use
fct_expand() to add this as a new level (recall that factors have a
fixed set of levels).
blank_country <- df %>%
filter(!country %in% high_four$country) %>%
mutate(country = fct_expand("")) %>%
mutate(country = factor("")) %>%
glimpse()
#> Observations: 136
#> Variables: 5
#> $ country <fct> , , , , , , , , , , , , , , , , , , , , , , , ,
#> $ continent <fct> Asia, Europe, Africa, Africa, Americas, Europe, Asia...
#> $ lifeExp <dbl> 43.828, 76.423, 72.301, 42.731, 75.320, 79.829, 75.6...
#> $ pop <int> 31889923, 3600523, 33333216, 12420476, 40301927, 819...
#> $ gdpPercap <dbl> 974.5803, 5937.0295, 6223.3675, 4797.2313, 12779.379...
# combine with the four countries
blank_country <- bind_rows(blank_country, high_four)
# check the values of country
unique(blank_country$country)
#> [1] "" "Canada" "Iceland" "Japan" "Reunion"Good. Ready to try geom_text_repel() as a new layer to the existing
graph. By trial and error, I edited some of the arguments to obtain a
text alignment I liked.
p3 <- p1 + ggrepel::geom_text_repel(data = blank_country,
aes(x = gdpPercap/10000, y = lifeExp, label = country),
color = rcb("dark_BG"),
box.padding = 0.5,
point.padding = 0.3,
min.segment.length = 0.4,
nudge_x = -0.1,
nudge_y = -5)
p3
I create a new data frame with one row that contains variables to locate
the text in one panel. The variable names that place the text have to be
the same columns as in the graph data frame, e.g., continent,
gdpPercap, and lifeExp. The new variable is label, which contains
the text I want to add to the panel
continent <- "Europe"
gdpPercap <- 0.5
lifeExp <- 60
label <- "Europe has the highest overall life expectancy\nand the smallest range from low to high of any world region."
# create the tibble
text_eu <- tibble(continent, gdpPercap, lifeExp, label) %>%
glimpse()
#> Observations: 1
#> Variables: 4
#> $ continent <chr> "Europe"
#> $ gdpPercap <dbl> 0.5
#> $ lifeExp <dbl> 60
#> $ label <chr> "Europe has the highest overall life expectancy\nand...Use geom_text() with the new data frame as the data argument.
p4 <- p3 + geom_text(data = text_eu,
mapping = aes(x = gdpPercap, y = lifeExp, label = label),
color = rcb("dark_BG"),
hjust = 0)
p4
Create a tibble like the one above, but this one has multiple rows; one row per text label.
text_panels <- wrapr::build_frame(
"continent" , "gdpPercap" , "lifeExp", "label" |
"Europe" , 0.5 , 60,
"Europe has the highest overall life expectancy\nand the smallest range from low to high of any world region." |
"Africa" , 1.5 , 70,
"Low GDP does not necessarily equate to\nlow life expectancy.")
glimpse(text_panels)
#> Observations: 2
#> Variables: 4
#> $ continent <chr> "Europe", "Africa"
#> $ gdpPercap <dbl> 0.5, 1.5
#> $ lifeExp <dbl> 60, 70
#> $ label <chr> "Europe has the highest overall life expectancy\nand...The geom_text() is identical to the one above excpet a change in data
in the geom.
p5 <- p3 + geom_text(data = text_panels,
mapping = aes(x = gdpPercap, y = lifeExp, label = label),
color = rcb("dark_BG"),
hjust = 0)
p5
Wickham H and Grolemund G (2017) R for Data Science. O’Reilly Media, Inc., Sebastopol, CA https://r4ds.had.co.nz/