Chapter 14

Scales and guides

This chapter has no exercises (yet!), so we show some examples to demonstrate the customization of scales.

library(tidyverse)
library(scales)
library(ggnewscale)
library(gt)
library(gtExtras)
library(cowplot)

A dummy data set to use for demonstrating scales in ggplot2 is shown in Figure 1

tb <- tibble(
  v_random_number = sample(0:100, size = 10, replace = FALSE),
  v_continuous = 1:10,
  v_discrete = sample(letters, 10),
  v_expo = (1:10)^4
)

tb |> gt() |> gt_theme_538()

v_random_number	v_continuous	v_discrete	v_expo
6	1	i	1
89	2	m	16
45	3	v	81
100	4	c	256
12	5	e	625
84	6	n	1296
2	7	u	2401
59	8	x	4096
55	9	w	6561
61	10	r	10000

Figure 1: Dummy data set created for demonstration of scales in ggplot2

14.1 Theory of scales and guides

Scale specification, naming scheme and fundamental scale types are shown in Figure 2.

g <- tb |> 
  ggplot(
    aes(
      x = v_continuous,
      y = v_random_number,
      colour = v_discrete
    )
  ) +
  geom_point(size = 4) +
  paletteer::scale_color_paletteer_d(
    "palettetown::croconaw"
  ) +
  labs(
    subtitle = "ggplot2 automatically assigns scale_*_continuous() to X and Y axis in this plot.\nThe colour scale becomes scale_colour_discrete()"
  ) +
  theme_cowplot() +
  theme(
    legend.position = "bottom"
  )
g

Figure 2: The demonstration of Scale specification, naming scheme and fundamental scale types in ggplot2

14.2 Scale names

The scales can be easily names, wither in the main scale_*_*() function using the argument name = "..." , or, more easily using the labs() helper function as shown below in Figure 3

g +
  labs(
    title = "A nice graph with a title",
    x = "A continuous variable from 1 to 10",
    y = "A number (1 to 100)",
    colour = "A discrete variable",
    subtitle = "The labs() function helps to easily add titles, subtites, legend titles and axes titles."
  )

Figure 3: Using labs() helper function to add a title to legends and axis

14.3 Scale breaks

The breaks = argument to the scale_*_*() functions is useful in determining the axis-ticks and lines or the colour levels in the legend to be displayed, as shown below in Figure 4

g1 <- tb |> 
  ggplot(aes(x = v_continuous,
             y = v_expo,
             colour = v_discrete)) + 
  geom_point(
    size = 6
    ) +
  scale_x_continuous(
    breaks = seq(0, 10, 2),
    name = "Breaks at even numbers"
  ) +
  scale_y_continuous(
    name = "Exponentially increasing numbers\nBreaks at interval of 2000",
    breaks = seq(0, 10000, 2000)
  ) +
  labs(
    title = "Using breaks argument in scale_*_* functions",
    subtitle = "This plot shows customized breaks at interval of 2 on x-axis and 1000 on y-axis"
  ) +
  theme_cowplot() +
  theme(legend.position = "none")
g1

Figure 4: Using breaks argument in scale_*_* functions to show customized intervals of 2 on x-axis, and, 1000 on y-axis

14.4 Scale limits

The Figure 5 shows an example of setting scale limits using the limit = argument in scale_*_*() functions in ggplot2. In this example, Y-axis limited upto 8000; and colour scale limtied to levels between a to h only.

g2 <- g1 +
  scale_y_continuous(
    limits = c(0, 8000),
    name = "Limits from 0 to 8000"
  ) +
  scale_colour_discrete(
    limits = letters[1:8],
    na.value = "lightgrey",
    name = "Limited to characters between a to h"
  ) +
  labs(
    title = "Setting scale limits",
    subtitle = "Y-axis limited upto 8000; and colour scale limtied to a - h only"
  ) +
  theme(legend.position = "bottom")
g2

Figure 5: Setting scale limits using limit argument in scale_*_* functions: Y-axis limited upto 8000; and colour scale limtied to some levels only

Further, with the oob = argument we can control the behavior of out-of-bounds data, as shown in Figure 6 below to squish to data onto the nearest limits.

g2 +
  scale_y_continuous(
    limits = c(0, 8000),
    name = "Limits from 0 to 8000; with squish oob",
    oob = squish
  ) +
  labs(
    subtitle = "Y-axis limited upto 8000, but with out of bounds value squished to 8000"
  )

Figure 6: Using oob argument to limit Y-axis up-to 8000, but with out of bounds value squished to 8000

14.5 Scale guides

To demonstrate the use of guides() function, we customize the colour legend from Figure 4 to generate a customized legend in Figure 7

g4 <- g1 +
  theme(legend.position = "bottom") +
  labs(
    title = "Customized Legend", 
    colour = "Using guide_legend() to customize legend") +
  guides(
    colour = guide_legend(
      direction = "vertical",
      reverse = TRUE,
      nrow = 2,
      override.aes = list(
        size = 4
      ),
      theme = theme(
        legend.text = element_text(
          face = "bold"
        )
      )
    )
  )
g4

Figure 7: Using guides() with guide_legend() to customize legend

14.6 Scale transformation

The continuous scales in Figure 7 can be transformed using the trans argument in the scale_*_*() function, as shown in Figure 8 below.

g4 + 
  scale_y_continuous(
    trans = "log2",
    name = "Transformed Y-Axis\nwith powers of Two"
  ) +
  labs(subtitle = "Transformed Y-Axis to the powers of 2",
       title = "Transformed Axes")

Figure 8: Continuous scale transformation into a log2 scale using trans argument in the scale_*_*() function

14.7 Legend merging and splitting

The Figure 9 shows us that if we’ve mapped colour to both points and lines, the keys will show both points and lines.

g4 +
  geom_col(
    fill = "transparent"
  ) +
  labs(subtitle = "Merged Legends - default option in ggplot2",
       title = "Legend merging and splitting (1)",
       colour = "Merged Legends")

Figure 9: Merged legends in ggplot2 - the default behaviour

The Figure 10 shows us how we can split or hide legends for certain geoms - here we-ve hidden the colour legend for lines / column bars - so the key will show only points (and not lines).

g4 +
  geom_col(
    fill = "transparent",
    show.legend = FALSE
  ) +
  labs(subtitle = "Splitting / Hiding Legends using show.legend = FALSE in one of the geoms",
       title = "Legend merging and splitting (2)",
       colour = "Legend only for points")

Figure 10: Hiding specific legends using show.legend = FALSE in one of the geoms

14.8 Legend key glyphs

Using the same graph as the last section, the Figure 11 demonstrates the use of argument key_glyph = within a geom_*() function to change the glyph used in the key of the graph.

g4 +
  geom_col(
    fill = "transparent",
    key_glyph = draw_key_path
  ) +
  labs(title = "Changing the glyphs for legends",
       colour = "Customized legend glyph")

Figure 11: The use of argument key_glyph with a geom_*() to change the glyph used in the key