facetious

facetious is home to some alternate facetting options for ggplot2.

What’s in the box

facet_wrap_strict()
- respects nrow, ncol and inserts empty grobs to keep the resultant plot at the given size
- Otherwise works like ggplot2::facet_wrap()
facet_grid_blank()
- When a factor level doesn’t exist, this facetting approach will insert a totally empty object, instead of just an empty plot
- Otherwise works like ggplot2::facet_grid()

Installation

You can install from GitHub with:

# install.package('remotes')
remotes::install_github('coolbutuseless/facetious')

`facet_wrap_strict`

In ggplot2::facet_wrap() (regardless of the nrow and ncol specified by the user), rows and columns of empty facets are removed.

This behaviour of ggplot2 can make it difficult to size and align plots in multiple plots.

In contrast, facetious::facet_wrap_strict() will strictly adhere to the nrow, ncol specified by the user, and retain blank rows/cols of facets.

library(ggplot2)
library(patchwork)
library(facetious)

#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# Standard ggplot2 facet_wrap()
#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
ggplot_facet_wrap <- ggplot(mtcars) +
  geom_point(aes(mpg, wt)) + 
  facet_wrap(~cyl, nrow = 3, ncol = 2) + 
  labs(title = "facet_wrap(... nrow=3, ncol=2)")

#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# facetious::facet_wrap_strict
#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
facetious_facet_wrap_strict <- ggplot(mtcars) +
  geom_point(aes(mpg, wt)) + 
  facet_wrap_strict(~cyl, nrow = 3, ncol = 2) + 
  labs(title = "facet_wrap_strict(... nrow=3, ncol=2)")

#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# Use 'patchwork' to stitch plots side-by-side
#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
ggplot_facet_wrap + facetious_facet_wrap_strict

`facet_grid_blank`

By default when a particular factor level is empty, ggplot2 includes an empty plot.

facetious::facet_grid_blank() goes a step further and (by default) makes the empty facet entirely blank.

That is, empty factor levels are represented as a grid::nullGrob(), but it is possible to specify another grob, or list of grobs to fill the empty facets.

v0.1.1 of facetious borrows an idea from the really great hack provided in ggbillboard by Duncan Garmonsway - i.e. don’t just put empty grobs into empty facets, put some advertising in there!

library(dplyr)
library(ggplot2)
library(patchwork)
library(facetious)

#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# Make some data with some empty factors
#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
plot_df <- mtcars %>%
  mutate(
    cyl = as.factor(cyl),
    am = as.factor(am)
  ) %>%
  filter(!(cyl == 4 & am == 1)) %>%
  filter(!(cyl == 8 & am == 0))

#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# Standard ggplot facet_grid()
#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
ggplot_facet_grid <- ggplot(plot_df) +
  geom_point(aes(mpg, wt)) +
  facet_grid(rows = vars(cyl), cols = vars(am), drop = FALSE,
             labeller = label_both) +
  labs(title = "ggplot2::facet_grid()")

#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# facetious::facet_grid_blank()
#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
facetious_facet_grid_blank <- ggplot(plot_df) +
  geom_point(aes(mpg, wt)) +
  facet_grid_blank(rows = vars(cyl), cols = vars(am), drop = FALSE, 
                   labeller = label_both) +
  labs(title = "facetious::facet_grid_blank()")

#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# Use 'patchwork' to stitch plots side-by-side
#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
ggplot_facet_grid + facetious_facet_grid_blank

Using alternate globs for blank facets. Example 1 - single grob

grob_for_blank <- grid::textGrob("No data available")

 ggplot(plot_df) +
  geom_point(aes(mpg, wt)) +
  facet_grid_blank(rows = vars(cyl), cols = vars(am), drop = FALSE, 
                   labeller = label_both, blank = grob_for_blank) +
  labs(title = "facetious::facet_grid_blank()")

Using alternate globs for blank facets. Example 2 - multiple grobs

grobs_for_blank <- list(
  grid::rasterGrob(jpeg::readJPEG("man/figures/distracted.jpg")),
  grid::rasterGrob(jpeg::readJPEG("man/figures/run.jpg"))
)

 ggplot(plot_df) +
  geom_point(aes(mpg, wt)) +
  facet_grid_blank(rows = vars(cyl), cols = vars(am), drop = FALSE, 
                   labeller = label_both, blank = grobs_for_blank) +
  labs(title = "facetious::facet_grid_blank()")

Related Software

ggplot2
ggforce a great package with alternate facets and tools for ggplot2
ggbillboard a package for replacing blank facets with other grobs. This was so cool that I’ve incorporated the idea into facetious as part of the facet process, rather than as a post-processing step.

Acknowledgements

R Core for developing and maintaining the language.
CRAN maintainers, for patiently shepherding packages onto CRAN and maintaining the repository

coolbutuseless / facetious