Customizing Plots

scales, labels, facet_wrap()

Author

Emily Malcolm-White 

library(tidyverse)
#Import the can_lang dataset 
can_lang <- read.csv("https://raw.githubusercontent.com/ttimbers/canlang/master/inst/extdata/can_lang.csv")

A starting graph: scatterplot of can_lang

can_lang_plot <- ggplot(can_lang, aes(x=most_at_home, y=mother_tongue)) +
  geom_point() + 
  xlab("Language spoken most at home \n (number of Canadian residents)") +
  ylab("Mother tongue \n (number of Canadian residents)")

Notice anything weird about this plot?

Axis display format: scales package

 # Install the package if needed
library(scales)

We want to customize how the continuous x and y axes look, so we need to use the argument labels=label_comma() inside a scale_*_continuous() layer:

can_lang_plot +
  scale_x_continuous(labels = label_comma()) +
  scale_y_continuous(labels = label_comma())
1
numbers on the x-axis are displayed with commas (and not in scientific notation)
2
numbers on the y-axis are displayed with commas (and not in scientific notation)

What other formats are available in the scales package?

When passing a formatting function inside scale_*_continuous(labels = ...) you have options!

Function Use Case Example Input Example Output
label_comma() Formats numbers with commas 1234567 "1,234,567"
label_dollar() Formats numbers as dollar currency 99.99 "$99.99"
label_dollar(prefix = "€") Formats numbers as euro currency 99.99 "99.99€"
label_percent() Converts decimals to percent 0.25 "25%"
label_pvalue() Formats p-values 0.00005 "<0.0001"

Anything else?

Logarithmic Axes Transformations

Applying a Log Transformation

When you apply a log transformation to an axis (or both axes) in a plot, you convert values using a logarithmic scale instead of a linear scale. This means:

  • Instead of evenly spaced values (1, 2, 3, 4, …), a logarithmic scale spaces values exponentially (1, 10, 100, 1000, …).
  • The distance between ticks represents a multiplicative factor instead of an additive one.

See how much more clearly we can see all the points!

For you to do this yourself, you need to use scale_*_log10() instead of scale_*_continuous():

can_lang_plot +
  scale_x_log10(labels = label_comma()) +
  scale_y_log10(labels = label_comma())
1
converts x-axis to a log-scale
2
converts y-axis to a log-scale

Use ✅ scale_*_log10() instead of 🚫log(variable)

Using percents on a log scale

mutate to create new columns with percentage of Canadians who speak the language as their mother tongue:

can_lang <- can_lang %>%
  mutate(
    mother_tongue_percent = (mother_tongue / 35151728) * 100,
    most_at_home_percent = (most_at_home / 35151728) * 100
  )

Scatterplot with Percents and Colors

Create a scatterplot with most_at_home_percent and mother_tongue_percent. Vary the color and shape of the points depending on the category of language. You may need to adjust the position of the legend:

can_lang_percent_plot <- ggplot(can_lang, aes(x = most_at_home_percent,
                     y = mother_tongue_percent )) +
  geom_point(aes(color = category, shape=category), alpha=0.5) +
  xlab("Language spoken most at home \n (percentage of Canadian residents)") +
  ylab("Mother tongue \n (percentage of Canadian residents)") +
  theme(legend.position = "top", legend.direction = "vertical") +
  scale_x_log10(labels = comma) +
  scale_y_log10(labels = comma)

can_lang_percent_plot
1
Use most_at_home_percent as the x-axis
2
Use mother_tongue_percent as as the y-axis
3
vary the shape and the color based on the category of language. Note this is included in the aesthetics of the points. It also would have been okay to put these directly inside the global aesthetics (ggplot(aes(...))) so that these characteristics apply to any layers.
4
Adjusts the position of the legend

Labels

Adding text to a plot is one of the most common forms of annotation. Most plots will not benefit from adding text to every single observation on the plot, but labeling outliers and other important points is very useful.

A add label for each language in this dataset using geom_text(aes(label = language)):

can_lang_percent_plot  + 
  geom_text(aes(label=language), 
              nudge_x = 0.25, 
              nudge_y=0.25)

Yikes! This is way too much going on in one plot. A few options to try when this happens:

  • Decrease the font size of the labels (using the size= argument inside geom_text).
  • Use the ggrepel package to spread out the labels a bit more
  • Pick out only a subset of the points to label

Using ggrepel

Artwork by @allisonhorst
library(ggrepel)

can_lang_percent_plot + 
  geom_text_repel(aes(label=language), max.overlaps = Inf)

Subset the labels

Create a new column for the labels. Use case_when (or ifelse) to only use the official language names and not to put a label for other language categories.

can_lang <- can_lang %>% 
  mutate(official_languages = case_when(category == "Official languages" ~ language, TRUE ~ NA ))
# We need to redo the base plot with the new can_lang dataset with the new official_languages column in it 
can_lang_percent_plot <- ggplot(can_lang, aes(x = most_at_home_percent,  y = mother_tongue_percent)) +
  geom_point(aes(color = category, shape=category)) +
  xlab("Language spoken most at home \n (percentage of Canadian residents)") +
  ylab("Mother tongue \n (percentage of Canadian residents)") +
  theme(legend.position = "top", legend.direction = "vertical") + 
  scale_x_log10(labels = comma) +
  scale_y_log10(labels = comma)


can_lang_percent_plot + 
  geom_text_repel(aes(label=official_languages, min.segment.length=0, box.padding=1))

Facet Wrap

facet_wrap() is a function in the ggplot2 package that allows you to create a multi-panel plot showing a similar plot over different subsets of the data, usually different values of a categorical variable.

Create separate side-by-side plots for each different category of language.

can_lang_percent_plot + 
  facet_wrap(~category)