Skip to contents

Standardise 'ggplot2' geom width

Source: R/get-width.R
get_width.Rd

Standardise the width in 'ggplot2' geoms to appear visually consistent across plots with different numbers of categories, panel dimensions, and orientations.

This can be used in geoms such as ggplot2::geom_bar()/ggplot2::geom_col(), ggplot2::geom_boxplot(), ggplot2::geom_errorbar().

Usage

get_width(
  ...,
  n = NULL,
  n_dodge = NULL,
  orientation = c("x", "y"),
  equiwidth = NULL,
  panel_widths = NULL,
  panel_heights = NULL
)

Arguments

...

Must be empty. Forces all other arguments to be named and allows trailing commas.

n

Number of categories in the orientation aesthetic (i.e. "x" or "y"). For faceted plots, use the maximum n within a facet.

n_dodge

Number of dodge categories. Must match the number of groups in the fill or colour aesthetic when using position_dodge().

orientation

Orientation: "x" for vertical (width appearance equiwidth to panel width), "y" for horizontal (width appearance equiwidth to panel height).

equiwidth

Numeric. Scaling factor that controls the width appearance. A value of 1 is the default. Increase to make a wider appearance, and decrease to make a thinner appearance. If NULL, uses the value set by set_equiwidth(), falling back to 1.

panel_widths

A grid::unit object specifying the panel width. If NULL (default), uses the value set in the current theme.

panel_heights

A grid::unit object specifying the panel height. If NULL (default), uses the value set in the current theme.

Value

A numeric width value passed to the width argument of geom_bar(), geom_col(), or similar geoms.

See also

set_equiwidth()

Examples

library(ggplot2)
library(dplyr)
#> 
#> Attaching package: ‘dplyr’
#> The following objects are masked from ‘package:stats’:
#> 
#>     filter, lag
#> The following objects are masked from ‘package:base’:
#> 
#>     intersect, setdiff, setequal, union
library(patchwork)

set_theme(
  theme_grey() +
    theme(panel.widths  = rep(unit(75, "mm"), 2)) +
    theme(panel.heights = rep(unit(50, "mm"), 2))
)
set_equiwidth(1)

p1 <- mpg |>
  ggplot(aes(x = drv)) +
  geom_bar(
    width = get_width(n = 3),
    colour = "black",
    fill = "grey",
  )

p2 <- diamonds |>
  ggplot(aes(x = color)) +
  geom_bar(
    width = get_width(n = 7),
    colour = "black",
    fill = "grey",
  )

p3 <- diamonds |>
  ggplot(aes(y = color)) +
  geom_bar(
    width = get_width(n = 7, orientation = "y"),
    colour = "black",
    fill = "grey",
  )

p4 <- mpg |>
  ggplot(aes(x = drv, group = factor(cyl))) +
  geom_bar(
    position = position_dodge(preserve = "single"),
    width = get_width(n = 3, n_dodge = 4),
    colour = "black",
    fill = "grey",
  )

p1 + p2 + p3 + p4


d <- tibble::tibble(
  continent = c("Europe", "Europe", "Europe", "Europe", "Europe",
                "South America", "South America"),
  country   = c("AT", "DE", "DK", "ES", "PK", "TW", "BR"),
  value     = c(10L, 15L, 20L, 25L, 17L, 13L, 5L)
)

max_n <- d |>
  count(continent) |>
  pull(n) |>
  max()

d |>
  mutate(country = forcats::fct_rev(country)) |>
  ggplot(aes(y = country, x = value)) +
  geom_col(
    width = get_width(n = max_n, orientation = "y"),
    colour = "black",
    fill = "grey",
  ) +
  facet_wrap(~continent, scales = "free_y") +
  scale_y_discrete(continuous.limits = c(1, max_n)) +
  coord_cartesian(reverse = "y", clip = "off")


mpg |>
  ggplot(aes(x = drv)) +
  geom_bar(
    width = get_width(n = 3, panel_widths = unit(160, "mm")),
    colour = "black",
    fill = "grey",
  ) +
  theme(panel.widths = unit(160, "mm"))