The function we are showing off today is the patchwork() function from the patchwork package.
2 What is it for?
The patchwork() function lets the user stitch together sets of plots in various configurations using simple operators (+, /, and |). This is accomplished by combining various plot objects into a single “patchwork” object. The user can also create more advanced customized displays with the plot_layout and the plot_annotation options.
Patchwork is designed to work well with ggplot(), whereas base R can struggle if a user is trying to do complicated things to plots not created with base R.
A function with similar utility (though less flexibility) is grid.arrange() from the gridExtra package.
3 Examples
Code
# Setup chunkpenguins <-drop_na(penguins)# Plot 1p1 <-ggplot(penguins, aes(x = flipper_length_mm,y = body_mass_g, color = species)) +geom_point(alpha =0.7) +labs(title ="Body mass vs Flipper length", x ="Flipper length (mm)", y ="Body mass (g)") +theme(plot.title =element_text(size =10, face ="bold", hjust =0.5),axis.title.x =element_text(size =10),axis.title.y =element_text(size =10),legend.title =element_text(size =10),legend.text =element_text(size =8),legend.key.size =unit(0.75, "lines"))# Plot 2p2 <-ggplot(penguins, aes(x = bill_length_mm,y = body_mass_g, color = species)) +geom_point(alpha =0.7) +labs(title ="Body mass vs Bill length", x ="Bill length (mm)", y ="Body mass (g)") +theme(plot.title =element_text(size =10, face ="bold", hjust =0.5),axis.title.x =element_text(size =10),axis.title.y =element_text(size =10),legend.title =element_text(size =10),legend.text =element_text(size =8),legend.key.size =unit(0.75, "lines"))# Plot 3p3 <-ggplot(penguins, aes(x = bill_depth_mm,y = body_mass_g, color = species)) +geom_point(alpha =0.7) +labs(title ="Body mass vs Bill depth", x ="Bill depth (mm)", y ="Body mass (g)") +theme(plot.title =element_text(size =10, face ="bold", hjust =0.5),axis.title.x =element_text(size =10),axis.title.y =element_text(size =10),legend.title =element_text(size =10),legend.text =element_text(size =8),legend.key.size =unit(0.75, "lines"))# Plot 4p4 <-ggplot(penguins, aes(x = species, y = body_mass_g, fill = sex)) +geom_boxplot() +labs(title ="Body mass by Sex & Species", x ="Species", y ="Body mass (g)") +theme(plot.title =element_text(size =10, face ="bold", hjust =0.5),axis.title.x =element_text(size =10),axis.title.y =element_text(size =10),legend.title =element_text(size =10),legend.text =element_text(size =8),legend.key.size =unit(0.75, "lines"))# Plot 5p5 <-ggplot(penguins, aes(x = island, y = body_mass_g, fill = species)) +geom_violin(trim =FALSE) +labs(title ="Body mass by Island & Species",x ="Island", y ="Body mass (g)") +theme(plot.title =element_text(size =10, face ="bold", hjust =0.5),axis.title.x =element_text(size =10),axis.title.y =element_text(size =10),legend.title =element_text(size =10),legend.text =element_text(size =8),legend.key.size =unit(0.75, "lines"))
Below we combine four base R plots without using patchwork() or ggplot(). This example is included as a reference point.
Code
# 2 x 2 layoutpar(mfrow =c(2, 2))# Four base R plots to illustrate the 2 x 2 figureplot(penguins$body_mass_g, penguins$flipper_length_mm,main ="Plot 1: Scatter",xlab ="Body mass (g)", ylab ="Flipper length (mm)")boxplot(body_mass_g ~ species, data = penguins,main ="Plot 2: Boxplot",xlab ="Species", ylab ="Body mass (g)")boxplot(body_mass_g ~ sex, data = penguins,main ="Plot 3: Boxplot by sex",xlab ="Sex", ylab ="Body mass (g)")plot(density(na.omit(penguins$body_mass_g)),main ="Plot 4: Density",xlab ="Body mass (g)")
We used base R because the par() method is really finicky with ggplot(), and we couldn’t get it to work with the plots we outlined in the setup code chunk. We could accomplish something nicer than this with the grid.arrange() function from the gridExtra package, but we are not comparing patchwork() to grid.arrange() in this demo.
The rest of the examples are created using patchwork(). Here are four of our ggplot() plots stitched together in a basic configuration
There are two sets of plots - an upper set (first two rows) and lower set (final row)
The upper set takes up 2/3rds of the vertical space
It contains the first three plots and a designated final spot for the key
The lower set takes up the bottom third of the vertical space
It contains the fourth plot with a designated spot for the key
Code
# Inset one plot inside anotherp1 +inset_element(p4, left =0.0, bottom =0.6, right =0.5, top =1) +plot_layout(guides ="collect", axes ="collect") +plot_annotation(title ='Penguin Morphology')
Here we have inset our fourth plot into the corner of our first plot. This isn’t a particularly compelling example, but it may be useful at some point to have a reference shape/distribution/variable/something that can be pinned into a corner.
We can also customize the layout manually by giving specific input. Let’s get all 5 of our plots involved for these examples. Let’s add some tags as well, so we can refer to plots without mentioning the individual titles.
The patchwork() function will always default to trying to make a roughly square output, and it isn’t afraid to stretch or compress the user’s plots as needed. As long as the both the layout and the individual plots inside the layout are rectangles, the customization should go through without errors.
4 Is it helpful?
The patchwork() function makes organizing plots a bit easier even than grid.arrange(). It is more customizable than most other options, and works well with ggplot(). It’s probably not the best thing since sliced bread, but it is a great tool generally for visualizing data; we think it is worth having in your toolkit.
