signs

library(scales)
library(signs)
library(dplyr)
library(ggplot2)
library(ggrepel)

The Basics

Using signs is simple, especially if you’re familiar with functions like number(), number_format(), comma(), comma_format(), percent(), and percent_format() from the scales package. It simply provides two new functions to complement these: signs() and signs_format().

  • Points in group 1 are labeled with a true Unicode minus glyph.
  • Points in group 2 are labeled with the traditional ASCII hyphen-minus.
  • Basic usage is identical.

Other Number Formats

You can use any formatting function with signs::signs()—not just scales::number(). Let’s assume everyone gets 8 hours of sleep, so we can label the points as percentages.

p +
  ylab("Extra Sleep (% over 8 hours)") +
  label_hours(
    aes(
      label = case_when(
        group == 1 ~ signs(extra / 8, accuracy = .1, format = scales::percent),
        group == 2 ~ percent(extra / 8, accuracy = .1)
      )
    )
  )

Or we can muliply by days in a year and use scales::comma().

p +
  ylab("Extra Sleep (hours / year)") +
  label_hours(
    aes(
      label = case_when(
        group == 1 ~ signs(extra * 365, format = scales::comma),
        group == 2 ~ comma(extra * 365)
      )
    )
  )

format can be any function that takes a numeric vector and returns a character vector.

Matching by Position

Note that format and all other options (see below) come after the dots; this way you can enjoy the same matching by position you know from scales. For example, if you prefer the simplicity of number(x, 1), you can simply use signs(x, 1).

x <- seq(-4, 4)

number(x, 1) # first argument is accuracy
#> [1] "-4" "-3" "-2" "-1" "0"  "1"  "2"  "3"  "4"
signs(x, 1)  # first argument is accuracy
#> [1] "−4" "−3" "−2" "−1" "0"  "1"  "2"  "3"  "4"

Other Arguments

signs::signs() offers 3 other arguments for convenience:

  1. add_plusses
  2. trim_leading_zeros
  3. label_at_zero

(#3 is addressed below under Axis Labels.)

Sometimes, as with this dataset, you want to show change from a baseline. You might not only want to include a − in front of negative numbers, but a + in front of positive numbers as well. This is as simple as add_plusses = TRUE.

p +
  label_hours(
    aes(
      label = case_when(
        group == 1 ~ signs(extra, accuracy = .1, add_plusses = TRUE),
        group == 2 ~ number(extra, accuracy = .1)
      )
    )
  )

If all values are on the interval (−1, 1), it can be more compact to remove leading zeros. Do this with trim_leading_zeros = TRUE:

Axis Labels

You can also use Unicode minus signs on an entire axis. This function is called signs::signs_format(), by analogy to scales::number_format(), scales::percent_format(), and the rest of the _format() functions. Note that it accepts the same optional arguments as signs::signs() as well.

You may want to treat zero itself differently, particularly when every other value has either a plus or a minus. Maybe you’ll be extra pedantic about it with label_at_zero = "symbol" (notice the y-axis labels below, not the data point labels):

Or, especially if the location of zero is already obvious, you might want to leave it blank with label_at_zero = "blank":

Setting Options Globally

You can set format, add_plusses, trim_leading_zeros, or label_at_zero globally for a script with options():

The defaults are scales::number, FALSE, FALSE, and "none", respectively.