Facets

Improvement over time?

Has fuel economy changed considerably between 1999 and 2008? Perhaps there is a difference if you also consider the car class? Experiment with facet_wrap(), facet_grid(), aesthetic mappings, and smoothing layers.

ggplot(_____) +
  geom_point() +
  facet_wrap(~___)

ggplot(_____) +
  geom_point() +
  geom_smooth() + 
  facet_grid(___ ~ ___)

► Solution:

We can split the simple scatterplot over two facets, one per year:

ggplot(data = mpg) +
  geom_point(mapping = aes(x = displ, y = hwy)) +
  facet_wrap(~year)

The smoothing layer seems to confirm a slight improvement, especially for engines with a displacement of three or more liters:

ggplot(data = mpg) +
  geom_point(mapping = aes(x = displ, y = hwy)) +
  geom_smooth(mapping = aes(x = displ, y = hwy)) +
  facet_wrap(~year)

## `geom_smooth()` using method = 'loess' and formula 'y ~ x'

We can also look at each car class in parallel.

ggplot(data = mpg) +
  geom_point(mapping = aes(x = displ, y = hwy)) +
  facet_grid(year~class)

But too many facets may be not as helpful, we can also use color:

ggplot(data = mpg) +
  geom_point(
    mapping = aes(x = displ, y = hwy, color = factor(year))
  ) +
  geom_smooth(
    mapping = aes(x = displ, y = hwy, color = factor(year)),
    method = "lm"
  ) +
  facet_wrap(~class)

Labeling the facets

What changes if you add the argument labeller = "label_both" to the facet_wrap() call?

► Solution:

This is useful if the value itself is not self-explanatory:

ggplot(data = mpg) +
  geom_point(mapping = aes(x = displ, y = hwy)) +
  facet_wrap(~cyl, labeller = "label_both")

Different scales

Is it possible to use a different x and y scale for each facet? How?

► Solution:

Via the scales argument, zooms in to the range of the corresponding scale(s).

ggplot(data = mpg) +
  geom_point(mapping = aes(x = displ, y = hwy)) +
  facet_wrap(
    ~cyl,
    labeller = "label_both",
    scales = "free_x"
  )

ggplot(data = mpg) +
  geom_point(mapping = aes(x = displ, y = hwy)) +
  facet_wrap(
    ~cyl,
    labeller = "label_both",
    scales = "free_y"
  )

ggplot(data = mpg) +
  geom_point(mapping = aes(x = displ, y = hwy)) +
  facet_wrap(
    ~cyl,
    labeller = "label_both",
    scales = "free"
  )

Other arguments

Experiment with other arguments to facet_wrap() and facet_grid().

More exercises

Find more exercises in Section 3.5.1 of r4ds.

Copyright © 2019 Kirill Müller. Licensed under CC BY-NC 4.0.