4 Making Your Own Theme
In 2018, BBC data journalists Nassos Stylianou and Clara Guibourg, along with their team, developed a custom ggplot theme that matches the BBC’s style. By introducing this bbplot package for others to use, they changed their organization’s culture, removed bottlenecks, and allowed the BBC to visualize data more creatively.
To understand the significance of these changes, it’s helpful to know how things worked at the BBC before the introduction of bbplot. In the mid-2010s, journalists who wanted to make data visualization had two choices:
They could use an internal tool. This tool could create data visualizations but was limited to the predefined charts it had been designed to generate.
They could use Excel to create mockups and then work with a graphic designer to finalize the charts. This approach led to better results, and was way more flexible, but required extensive, time-consuming back-and-forth with a designer.
Neither of these choices was ideal, and they limited the BBC’s data visualization output. R freed the journalists from having to work with a designer. It wasn’t that the designers were bad (they weren’t), but ggplot allowed the journalists to explore different visualizations on their own. Working with a designer required the journalists to have a fully-formed idea that the designer could take and improve upon.
As the team improved their ggplot skills, they realized that it might be possible to produce more than just exploratory data visualizations. Could they create production-ready charts in R that could go straight onto the BBC website? In this chapter, I discuss the power of custom ggplot themes. I then go through the code in the bbplot package to learn how custom themes work. I wrap up the chapter by exploring the impact that bbplot had, not only on a technical level, but also as a catalyst for a larger culture change.
The Power of a Theme
As Stylianou, Guibourg, and their colleagues realized, so much of the work involved in making a professional chart consists of small tweaks. What font should you use? Where should the legend go? Should axes have titles? Should charts have grid lines? These questions may seem small, but they have a big impact on the final product.
And these are the types of questions to which a custom theme can provide answers. Custom themes force everyone who uses them follows style guidelines and ensures that all data visualization is on brand. What’s more, when more experienced R users in an organization make a custom theme, other less experienced users can take advantage of their work to make sure their plots follow organizational style guidelines. Custom themes, as we’ll see below when we dive into the bbplot code, involve a set of code that makes a set of small tweaks to all plots. Rather than forcing everyone to copy the long code to tweak each plot they make, putting this code into a custom theme allows everyone to apply the theme with one line of code.
Using bbplot to Style a Penguin Plot
The bbplot package has two functions: bbc_style() and finalise_plot(). The latter deals with things like adding the BBC logo, saving plots in the correct dimensions, and other tasks done after the plot is complete (we’ll discuss it a bit later on). For now, let’s look at the bbc_style() function, which applies a custom ggplot theme to any plot, making all plots look consistent and follow BBC style guidelines.
Creating an Example Plot
To show how this function works, let’s create a plot. We’ll do so using the palmerpenguins package, which has data on penguins living on three islands in Antarctica. To give you a sense of what this data looks like, let’s load the palmerpenguins and tidyverse packages.
We now have data that we can work with in an object called penguins. Here’s what the first ten rows look like.
#> # A tibble: 344 × 8
#> species island bill_length_mm bill_depth_mm
#> <fct> <fct> <dbl> <dbl>
#> 1 Adelie Torgersen 39.1 18.7
#> 2 Adelie Torgersen 39.5 17.4
#> 3 Adelie Torgersen 40.3 18
#> 4 Adelie Torgersen NA NA
#> 5 Adelie Torgersen 36.7 19.3
#> 6 Adelie Torgersen 39.3 20.6
#> 7 Adelie Torgersen 38.9 17.8
#> 8 Adelie Torgersen 39.2 19.6
#> 9 Adelie Torgersen 34.1 18.1
#> 10 Adelie Torgersen 42 20.2
#> # ℹ 334 more rows
#> # ℹ 4 more variables: flipper_length_mm <int>,
#> # body_mass_g <int>, sex <fct>, year <int>To get our data in a more usable format, let’s count how many penguins live on each island. We do this with the count() function from the dplyr package (one of several packages that are loaded when we load the tidyverse). This gives us some simple data that we can use for plotting:
This gives us some simple data that we can use for plotting below.
#> # A tibble: 3 × 2
#> island n
#> <fct> <int>
#> 1 Biscoe 168
#> 2 Dream 124
#> 3 Torgersen 52Because we’re going to use this data multiple times below, let’s save it as an object called penguins_summary.
Now that we’ve got some data to work with, we’re ready to create a plot. Before showing what bbplot does, let’s make our plot with the ggplot defaults. Here is the code we’ll use:
We use our penguins_summary data frame, putting the island on the x axis and the count of the number of penguins (n) on the y axis, and making each bar a different color with the fill aesthetic property. We’ll modify this plot multiple times, so to simplify this process, we save it as an object called penguins_plot. The resulting plot is seen in Figure 4.1.
Figure 4.1: A chart with the default theme
It isn’t the most aesthetically pleasing chart. The gray background is ugly, the y axis title is hard to read because it’s angled, and the text size overall is quite small. But don’t worry: we’ll be improving it soon!
Applying the bbc_style() Function
Now that we have a basic plot to work with, let’s make it look like a BBC chart. To do this, we load the bbplot package:
library(bbplot)We can then apply the bbc_style() function to our penguins_plot.
penguins_plot +
bbc_style()Take a look at what happens in Figure 4.2 with the application of bbc_style() to our plot.
Figure 4.2: The same chart with BBC style
Way different, right? Larger font size, legend on top, no axis titles, stripped down grid lines, and a white background. These are the major changes that the bbc_style() function makes. Let’s look at them one by one.
Breaking Down the Custom Theme
Here is the code for the bbc_style() function (taken from the bbplot GitHub repository at https://github.com/bbc/bbplot, with some minor tweaks for readability). The first line gives the function a name and indicates that what follows is, in fact, a function definition. We’ll discuss functions more in Chapter ??.
bbc_style <- function() {
font <- "Helvetica"
ggplot2::theme(
# TEXT FORMAT
# This sets the font, size, type and colour
# of text for the chart's title
plot.title = ggplot2::element_text(
family = font,
size = 28,
face = "bold",
color = "#222222"
),
# This sets the font, size, type and colour
# of text for the chart's subtitle,
# as well as setting a margin between the title and the subtitle
plot.subtitle = ggplot2::element_text(
family = font,
size = 22,
margin = ggplot2::margin(9, 0, 9, 0)
),
# This leaves the caption text element empty,
# because it is set elsewhere in the finalise plot function
plot.caption = ggplot2::element_blank(),
# LEGEND FORMAT
# This sets the position and alignment of the legend,
# removes a title and background for it
# and sets the requirements for any text within the legend.
# The legend may often need some more manual tweaking
# when it comes to its exact position based on the plot coordinates.
legend.position = "top",
legend.text.align = 0,
legend.background = ggplot2::element_blank(),
legend.title = ggplot2::element_blank(),
legend.key = ggplot2::element_blank(),
legend.text = ggplot2::element_text(
family = font,
size = 18,
color = "#222222"
),
# AXIS FORMAT
# This sets the text font, size and colour for the axis test,
# as well as setting the margins and removes lines and ticks.
# In some cases, axis lines and axis ticks are things we would
# want to have in the chart -
# the cookbook shows examples of how to do so.
axis.title = ggplot2::element_blank(),
axis.text = ggplot2::element_text(
family = font,
size = 18,
color = "#222222"
),
axis.text.x = ggplot2::element_text(margin = ggplot2::margin(5, b = 10)),
axis.ticks = ggplot2::element_blank(),
axis.line = ggplot2::element_blank(),
# GRID LINES
# This removes all minor gridlines and adds major y gridlines.
# In many cases you will want to change this to remove
# y gridlines and add x gridlines.
# The cookbook shows you examples for doing so.
panel.grid.minor = ggplot2::element_blank(),
panel.grid.major.y = ggplot2::element_line(color = "#cbcbcb"),
panel.grid.major.x = ggplot2::element_blank(),
# BLANK BACKGROUND
# This sets the panel background as blank, removing the standard
# grey ggplot background colour from the plot.
panel.background = ggplot2::element_blank(),
# STRIP BACKGROUND
# This sets the panel background for facet-wrapped plots to white,
# removing the standard grey ggplot background colour and sets the
# title size of the facet-wrap title to font size 22.
strip.background = ggplot2::element_rect(fill = "white"),
strip.text = ggplot2::element_text(size = 22, hjust = 0)
)
}You’ll see that instead of loading the package ggplot2 with the code library(ggplot2) and then using the theme() function, the code below uses ggplot2::theme(). This indicates that the theme() function comes from the ggplot2 package. Writing code in this way is something that is done when making an R package, something we’ll discuss in Chapter ??.
Nearly all of the code in the bbc_style() function exists within the theme() function from ggplot2. Remember from Chapter 3 that theme() makes additional tweaks to an existing theme; it isn’t a complete theme like theme_light(), which will change the whole look-and-feel of your plot. In other words, by jumping straight into the theme() function, bbc_style() makes tweaks to the ggplot defaults.
As you can see, the bbc_style() function does a lot of tweaking. Let’s go through the changes it makes, section by section.
Text Formatting
The first code section formats the text. It defines a variable called font and assigns it the value Helvetica. This allows later sections to simply write font rather than repeating Helvetica over and over again. Also, if the BBC team ever wanted to use a different font, they could change Helvetica to, say, Comic Sans and update the font of all BBC plots (though I suspect higher-ups at the BBC might not be on board).
Subsequent pieces of this section make changes to the title, subtitle, and caption using the following pattern:
AREA_OF_CHART = ELEMENT_TYPE(
PROPERTY = VALUE
)We begin by selecting an area of the chart (for example, plot.title). Then, we say what type of element it is: element_text(), element_line(), element_rect(), or element_blank(). For now, we’re working with element_text() to handle formatting the title, subtitle, and caption. Within the element type, we give values to properties. This can be, say, setting the font family (the property) to Helvetica (the value).
One of the main things the bbc_style() function does is bump up the text size. Increasing font size helps with legibility, especially when plots made using the bbplot package are viewed on smaller mobile devices. The code first formats the title (with plot.title) using Helvetica 28-point bold font in a nearly black color (that’s the hex code #222222). The subtitle (using plot.subtitle) is 22-point Helvetica. Some spacing is added between the title and subtitle using the margin() function, which gives the spacing, in points, for the top (9), right (0), bottom (9), and left (0) sides. Finally, the caption (through the plot.caption argument) is removed using the element_blank() function. This is done because the finalise_plot() function in the bbplot package adds elements, including a caption and the BBC logo to the bottom of plots.
penguins_plot +
theme(
plot.title = element_text(
family = font,
size = 28,
face = "bold",
color = "#222222"
),
plot.subtitle = element_text(
family = font,
size = 22,
margin = margin(9, 0, 9, 0)
),
plot.caption = element_blank()
)We can see these changes in Figure 4.3.
Figure 4.3: Our chart with only text formatting changed
We then save our plot as an object in order to work with it in the next section.
penguins_plot_text <- penguins_plot +
theme(
plot.title = element_text(
family = font,
size = 28,
face = "bold",
color = "#222222"
),
plot.subtitle = element_text(
family = font,
size = 22,
margin = margin(9, 0, 9, 0)
),
plot.caption = element_blank()
)Legend Formatting
Next, we deal with the legend, putting it on top of the plot and left-aligning the text within it. Then, we remove the legend background (which would show up only if the background color of the entire plot were different), title, and legend key (the borders on the red, green, and blue boxes that show the island names). Finally, we make the legend’s text 18-point Helvetica with the same nearly black color.
penguins_plot_text +
theme(
legend.position = "top",
legend.text.align = 0,
legend.background = element_blank(),
legend.title = element_blank(),
legend.key = element_blank(),
legend.text = element_text(
family = font,
size = 18,
color = "#222222"
)
)We can see the result in Figure 4.4.
Figure 4.4: Our chart with changes to the legend
And again, we save this plot so we can continue to alter it below.
penguins_plot_legend <- penguins_plot_text +
theme(
legend.position = "top",
legend.text.align = 0,
legend.background = element_blank(),
legend.title = element_blank(),
legend.key = element_blank(),
legend.text = element_text(
family = font,
size = 18,
color = "#222222"
)
)Axis Formatting
Next are the axes. The code first removes axis titles because, as Nassos told me, these tend to take up a lot of chart real estate, and you can use the title and subtitle to make it clear what the axes show.
penguins_plot_legend +
theme(
axis.title = element_blank(),
axis.text = element_text(
family = font,
size = 18,
color = "#222222"
),
axis.text.x = element_text(margin = margin(5, b = 10)),
axis.ticks = element_blank(),
axis.line = element_blank()
)All text on axes becomes 18-point Helevetica and nearly black. The text on the x axis (in our case, Biscoe, Dream, and Torgersen) gets a bit of spacing around it. And, finally, we remove both axis ticks and axis lines. We can see the changes to our axes in Figure 4.5.
Figure 4.5: Our chart with changes to axis formatting
Let’s now save this plot as an object for future tweaks.
penguins_plot_axes <- penguins_plot_legend +
theme(
axis.title = element_blank(),
axis.text = element_text(
family = font,
size = 18,
color = "#222222"
),
axis.text.x = element_text(margin = margin(5, b = 10)),
axis.ticks = element_blank(),
axis.line = element_blank()
)Grid Lines Formatting
Now that we’ve tweaked overall text formatting, the legend, and the axes, let’s move onto grid lines. The approach here is fairly straightforward: remove all minor grid lines and the major grid lines on the x axis, keeping only major grid lines on the y axis, but making them a light gray (using the #cbcbcb hex code).
penguins_plot_axes +
theme(
panel.grid.minor = element_blank(),
panel.grid.major.y = element_line(color = "#cbcbcb"),
panel.grid.major.x = element_blank()
)We can see the result of these tweaks to the grid lines in Figure 4.6.
Figure 4.6: Our chart with tweaks to the grid lines
And, once again, we save our plot to an object.
Background Formatting
The previous iteration of our plot still had a gray background. The bbc_style() function removes this with the following code.
penguins_plot_grid_lines +
theme(
panel.background = element_blank()
)The plot without the gray background is seen in Figure 4.7.
Figure 4.7: Our chart with the gray background removed
There we go! We’ve recreated the Penguin plot using the bbc_style() function.
Small Multiples Formatting
However, you may recall that the function contains a bit more code, to modify strip.background and strip.text. These elements become relevant in small multiples charts like the one discussed in Chapter 2. Let’s turn our penguin chart into a small multiples chart to see these components of the BBC’s theme. I’ve used the code from the bbc_style() function, minus the sections that deal with small multiples, to make Figure 4.8.
Figure 4.8: Small multiples chart with no changes to the strip text formatting
When we use the facet_wrap() function, to make a small multiples chart, we are left with one chart per island. But note that, by default, the text above each chart is noticeably smaller than the rest of the chart. And the gray background behind the text stands out when we have removed the gray background from other parts of the chart. The consistency we’ve worked toward is now gone, with small text that is out of proportion to the other text in the chart and a gray background that sticks out like a sore thumb in a chart with an all white background.
I’ve saved the code used to make Figure 4.8 as an object, penguins_plot_weight. We now use this object in order to show how to change the text that shows up above each small multiples chart (called the strip in ggplot):
penguins_plot_weight +
theme(
strip.background = element_rect(fill = "white"),
strip.text = element_text(size = 17, hjust = 0, face = "bold")
)We remove the background (or, more accurately, color it white). Then we make the text larger, bold, and left aligned using hjust = 0. I did have to make the text size slightly smaller to fit in the book and added code to make it bold. You can see the result in Figure @ref(fig: penguins-plot-facetted-bbc-plot).
Figure 4.9: Small multiples chart in the BBC style
If you take a look at any chart on the BBC site, you’ll see how similar it is to our chart. All of the tweaks in the bbc_style() function (text formatting, legends, axes, grid lines, and backgrounds) that we used to make our example show up in charts seen by millions on the BBC website.
What About Colors?
You might be thinking: wait, what about the color of the bars? Doesn’t the theme change those? It’s a common point of confusion. If we read the documentation for the theme() function, though, it becomes clearer why this is the case:
Themes are a powerful way to customize the non-data components of your plots: i.e. titles, labels, fonts, background, gridlines, and legends.
Color (or, technically, in the case of the bar charts we have made in this chapter, fill) is used in plots as an aesthetic property to show something about data. In our small multiples chart, for instance, fill is mapped to the island (Biscoe is salmon, Dream is green, and Torgersen is blue). As we saw in Chapter 3, we can change fill using the various scale_fill_ functions. It is because fill is tied to the data rather than being about the overall look-and-feel that ggplot themes do not, on their own, change this component of plots.
In Conclusion: Code is the Catalyst for Culture Change
When Stylianou and Guibourg started developing a custom theme for the BBC, they had one question: would they be able to create graphs in R that could go directly onto the BBC website? And, wouldn’t you know, they succeeded! The bbplot package allowed them to make plots with a consistent look-and-feel that followed BBC standards and, most importantly, did not need help from a designer.
You can see many of the principles of high-quality data visualization discussed in Chapter 3 in this custom theme. In particular, the removal of extraneous elements (axis titles and grid lines, for instance) helps keep the focus on the data itself. And because applying the theme requires users to add only a single line to their ggplot code, it became simple to get others on board. Users had only to append bbc_style() to their code to produce a BBC-style plot.
Over time, others at the BBC noticed the data journalism team’s production-ready graphs and wanted to make their own. The team members set up R trainings for their colleagues and developed a “cookbook” (found at https://bbc.github.io/rcookbook/) that showed how to make various types of charts. Soon, the quality and quantity of BBC’s data visualization exploded. Stylianou told me, “I don’t think there’s been a day where someone at the BBC hasn’t used the package to produce a graphic.”
Now that you’ve seen how custom ggplot themes work, I hope you might be inspired to make one of your own. As you’ve seen, custom themes are a set of small tweaks that you can apply to plots to give them a consistent look-and-feel. Developing a custom theme can take your data visualization from meh to wow. And, once you’ve written the code, it only takes one line of code to apply your custom theme. If a custom theme can transform the data visualization work of the BBC, imagine what it can do for you.