Compare groups with facets · phiz

Split a scatterplot of car weight against fuel economy into one panel for each number of cylinders.

Data

Datasets::mtcars() returns the 32 cars of R’s mtcars data, bundled with phiz as resources/data/mtcars.csv. The example uses three of its columns:

model	mpg	cyl	wt
Mazda RX4	21.0	6	2.620
Mazda RX4 Wag	21.0	6	2.875
Datsun 710	22.8	4	2.320

wt is the weight in thousands of pounds, mpg is the fuel economy in miles per gallon, and cyl is the number of cylinders.

Code

use Phiz\Datasets;use function Phiz\{    aes, facet_wrap, geom_point, ggplot, labs,};$mtcars = Datasets::mtcars();$plot = ggplot($mtcars, aes(x: 'wt', y: 'mpg')) |>    geom_point() |>    facet_wrap('cyl', ncol: 3) |>    labs(x: 'Weight (1,000 lb)', y: 'Miles per gallon');$plot->save('facets.html');

use Phiz\Datasets;use Phiz\Plot;$mtcars = Datasets::mtcars();$plot = Plot::of($mtcars)    ->aes(x: 'wt', y: 'mpg')    ->point()    ->facetWrap('cyl', ncol: 3)    ->labs(x: 'Weight (1,000 lb)', y: 'Miles per gallon');$plot->save('facets.html');

Runnable script: facets.php (PHP 8.5+) or facets.chain.php (PHP 8.3–8.4).

Result

Vega-Lite specification

Shared scales. By default every panel has the same axis ranges, so positions compare across panels.
Free scales. scales: 'free', 'free_x', or 'free_y' gives each panel its own ranges. Positions then no longer compare across panels.
Mixed effects. Heavier cars also tend to have more cylinders. Each panel shows the weight and economy association among cars with the same cylinder count.
Small groups. The panels hold 11, 7, and 14 cars, so a trend within one panel is a rough description.

In R: ggplot(mtcars, aes(wt, mpg)) + geom_point() + facet_wrap(~cyl, ncol = 3).