Customizing colors

Not everyone has experience coding, but everyone has a favorite color. To help beginners feel a sense of ownership for the work they do, tmtyro prioritizes simple customization using a single function: change_colors(). It takes the place of twelve different scaling functions from ggplot2 while providing a standard interface for any visualization created by tmtyro.

Colors are tricky

ggplot2 can set color in many different ways, but only one way will work for any given scenario. Clearest is the distinction between color and fill. The first applies color to points, lines, and edges. It’s adjusted with functions that have color in their names, like scale_color_manual(). The fill aesthetic, on the other hand, defines areas, inside shapes, and in bars. It’s adjusted with functions that have fill in their names, like scale_fill_manual(). In every case, the _fill_ or _color_ part of a function will indicate its target.

Difficulty grows from there. Colors and fills are set in different ways for discrete data, continuous data, and binned data. Worse, color palettes are chosen using three incompatible methods depending on color set, with one method for custom manual palettes, another method for Brewer palettes, and a third for Viridis palettes. Many other options are available through additional packages, but this combination of two aesthetics, three types of data, and three types of palettes are built in to ggplot2.

	discrete data	numeric data
Nine functions for adjusting color depending on data type and palette choice
	discrete data	continuous	binned
manual¹	`scale_color_manual()`	`scale_color_gradient()`	`scale_color_steps()`
Brewer²	`scale_color_brewer()`	`scale_color_distiller()`	`scale_color_fermenter()`
Viridis³	`scale_color_viridis_d()`	`scale_color_viridis_c()`	`scale_color_viridis_b()`
¹ For manual palettes, define a vector of colors using the `values` argument. Set gradient colors in the `high` and `low` (and optionally `mid`) arguments. Colors can be chosen from a list of named colors or defined with a hex code
² For Brewer palettes, choose a numbered or named set with the `palette` argument.
³ For Viridis palettes, choose a lettered or named set with the `option` argument.

This table shows half of the 18 commonest functions for changing colors, but ggplot2 offers 44 without counting spelling variants. And each row of functions uses different parameters for choosing colors. The path to color customization is steep.

`change_colors()` is easy

All visualization functions from tmtyro can be changed from one standard method: change_colors(). This function considers a figure, figures out whether it makes more sense to change color or fill, and applies a standard interface for manual palettes, Brewer palettes, and Viridis palettes.

change_colors() manages differentiation among data types.

scale_color_manual() works only for discrete data, and scale_color_gradient() is only good with continuous values.
scale_color_brewer() will only work for discrete data types, while scale_color_distiller() works only for continuous data
scale_color_viridis_d() will only work for discrete data types, while scale_color_viridis_c() will only work for continous data.
change_colors() accommodates discrete and continuous data

change_colors() also introduces one standard interface of arguments.

scale_color_manual() sets colors using named or hexadecimal colors in the values argument
scale_color_brewer() sets colors using numbers with the palette argument
scale_color_viridis_d() sets colors using letters with the option argument
change_colors() uses the palettes argument for everything

When code is easy, the only difficult part is choice.

Options are many

manual colors

Most simply, change_colors() will set the colors you choose. Setting four colors for four items will assign them directly; any other number will make a gradient.

Color names like “pink” and “orange” work in R, as do specific hues like “forestgreen” and “steelblue.”¹ In addition to named colors, R will accept colors as “hex codes” using hexadecimal notation.² The first two digits of a hex code describe how red a color is from 0 to 255; the middle two describe how green it is; and the last two describe how blue. Combinations in the following chart give a sense of how they work, but an online color picker may help to narrow things down.³

Use manual colors—as named colors or as hex codes—by combining them in a vector inside change_colors():

dubliners_count |> 
  change_colors(c("#00BBBB", "tan", "purple"))

Custom colors also work well when naming particular values:

joyce_count |> 
  change_colors(c(
    "#88DD00", "#00DDDD",
    he = "red"))

Brewer palettes

If you’d rather not pick colors manually, Brewer palettes are an excellent choice. The Brewer qualitative palettes are well suited for discrete data, using color to distinguish categories like documents or words.

Brewer’s sequential palettes are ideal for showing differences in magnitude:

Choose a Brewer palette by using its name or number in the palette argument:

joyce_count |> 
  change_colors("Brewer", palette = "Dark2")

Viridis palettes

Viridis palettes offer another set of choices for colors in your visualizations. These palettes not only look beautiful on the screen, but they typically work well for monochrome print and are designed to accommodate most color vision needs.

These palettes work especially well for continuous data. Option “H” or “turbo” could work for discrete scales, but it also has a few caveats: among them, it’s poorly suited for black and white printing since it doesn’t follow a linear path between dark and light.

Choose a Viridis palette by using its name or letter in the palette argument: