3  Customising plots

3.1 Intended learning outcomes

By the end of this chapter you will be able to:

• Use scale_* functions to control axis labels, breaks and limits
• Distinguish coord_cartesian() from scale_* limits
• Apply built-in themes and set a global default with theme_set()
• Add titles, subtitles and captions with labs()
• Use the colour-blind-friendly viridis palettes

3.2 Functions used

• ggplot2: scale_x_discrete(), scale_x_continuous(), scale_y_continuous(), coord_cartesian(), scale_fill_viridis_d(), scale_colour_viridis_d(), labs(), theme_minimal(), theme_bw(), theme_classic(), theme_set()
• scales: label_comma() (replacement for scales::comma)

3.3 Scientific notation and number formatting

The column chart from sec-basic-plots uses scientific notation on the y-axis because the numbers are so large.

ggplot(port_summary, aes(x = start_port, y = embarked, fill = start_port)) +
  geom_col(show.legend = FALSE) +
  scale_fill_viridis_d(option = "E")

Figure 3.1: Total embarked, with the default y-axis in scientific notation.

We can override that with scale_y_continuous() and a label-formatting function from the scales package.

ggplot(port_summary, aes(x = start_port, y = embarked, fill = start_port)) +
  geom_col(show.legend = FALSE) +
  scale_fill_viridis_d(option = "E") +
  scale_y_continuous(labels = scales::label_comma())

Figure 3.2: Total embarked, with the y-axis formatted with thousands separators.

Note

scales::label_comma() is the modern factory function. The older scales::comma (no parentheses) still works but is being phased out.

3.4 Axis labels and titles

labs() is the most concise way to set axis labels, the plot title, subtitle and caption.

ggplot(port_summary, aes(x = start_port, y = embarked, fill = start_port)) +
  geom_col(show.legend = FALSE) +
  scale_fill_viridis_d(option = "E") +
  scale_y_continuous(labels = scales::label_comma()) +
  labs(x = "Port of origin",
       y = "Number embarked",
       title = "Number embarked on ships originating in British ports")

Figure 3.3: Adding axis labels and a title.

3.5 Themes

ggplot2 has several built-in themes. We have already set theme_minimal() as the global default in the project setup (see R/my_setup.R). You can override it for an individual plot by adding a different theme as a layer.

p <- ggplot(port_summary, aes(x = start_port, y = embarked, fill = start_port)) +
  geom_col(show.legend = FALSE) +
  scale_fill_viridis_d(option = "E") +
  scale_y_continuous(labels = scales::label_comma()) +
  labs(x = NULL, y = NULL)

(p + theme_minimal() + ggtitle("theme_minimal")) +
  (p + theme_bw() + ggtitle("theme_bw")) +
  (p + theme_classic() + ggtitle("theme_classic"))

Figure 3.4: The same plot in three different themes.

Tip

Type theme_ in the RStudio source pane and let auto-complete show you the available options. To set a theme for every subsequent plot in your script, run theme_set(theme_bw()) near the top.

3.6 Discrete x-axis labels

scale_x_discrete() lets you control the tick labels on a categorical axis without renaming the underlying data.

ggplot(port_summary, aes(x = start_port, y = embarked, fill = start_port)) +
  geom_col(show.legend = FALSE) +
  scale_x_discrete(labels = c("Liverpool", "Port of London", "Bristol")) +
  scale_fill_viridis_d(option = "E") +
  scale_y_continuous(labels = scales::label_comma())

Figure 3.5: Renaming x-axis ticks with scale_x_discrete().

3.7 Axis limits the right way

There are two ways to restrict the visible range of an axis, and they behave differently.

coord_cartesian(xlim = c(...), ylim = c(...)) zooms in on the plot. Data outside the range are still used in calculations (such as drawing a smoother) and are simply cropped from view.

scale_x_continuous(limits = c(...)) filters the data. Any rows outside the range are dropped before geoms or smoothers see them. This can silently change a regression line, so it is usually the wrong choice.

crop <- ggplot(dat_filter, aes(arrival_year, total_embarked)) +
  geom_point(alpha = 0.2) +
  geom_smooth() +
  coord_cartesian(xlim = c(1700, 1815)) +
  labs(title = "coord_cartesian (crop)")

filt <- ggplot(dat_filter, aes(arrival_year, total_embarked)) +
  geom_point(alpha = 0.2) +
  geom_smooth() +
  scale_x_continuous(limits = c(1700, 1815)) +
  labs(title = "scale_x_continuous(limits=)")

crop + filt

Figure 3.6: coord_cartesian (left) crops the view; scale limits (right) filter the data.

3.8 Viridis colour palettes

The default ggplot2 colours are not colour-blind friendly. The viridis family of palettes is and is now included in ggplot2. Use scale_fill_viridis_d() for discrete fill (a category mapped to fill), and scale_fill_viridis_c() for continuous fill (a number mapped to fill); equivalents exist for colour.

The option argument picks one of five palettes:

• "A" magma
• "B" inferno
• "C" plasma
• "D" viridis (the default)
• "E" cividis (the most accessible)

make_plot <- function(opt) {
  ggplot(port_summary, aes(start_port, embarked, fill = start_port)) +
    geom_col(show.legend = FALSE) +
    scale_fill_viridis_d(option = opt) +
    labs(title = paste("option =", opt), x = NULL, y = NULL)
}

make_plot("A") + make_plot("B") + make_plot("C") +
  make_plot("D") + make_plot("E") +
  plot_layout(nrow = 2)

Figure 3.7: Five viridis palettes applied to the same column chart.

3.9 Activities

TipActivity 1

Apply the theme_classic() theme to the column chart of port totals and edit the y-axis label to “Total embarked (people)”.

NoteSolution

ggplot(port_summary, aes(x = start_port, y = embarked, fill = start_port)) +
  geom_col(show.legend = FALSE) +
  scale_fill_viridis_d(option = "E") +
  scale_y_continuous(labels = scales::label_comma()) +
  labs(x = "Port of origin", y = "Total embarked (people)") +
  theme_classic()

TipActivity 2

Replace scale_x_continuous(limits = c(1700, 1815)) in the right-hand panel above with coord_cartesian(xlim = c(1700, 1815)). Confirm that the smoother shape is the same as the left panel.