I often see bar charts where the bars are directly labeled with the value they represent. In this post I will walk you through how you can create such labeled bar charts using
The data I will use comes from the 2019 Stackoverflow Developer Survey. To make creating the plot easier I will use the
bar_chart() function from my
ggcharts package which outputs a
ggplot that can be customized further using any
library(dplyr) library(ggplot2) library(ggcharts) dreaded_lang <- tibble::tribble( ~language, ~pct, "VBA", 75.2, "Objective-C", 68.7, "Assembly", 64.4, "C", 57.5, "PHP", 54.2, "Erlang", 52.6, "Ruby", 49.7, "R", 48.3, "C++", 48.0, "Java", 46.6 ) chart <- dreaded_lang %>% bar_chart(language, pct) %>% print()
To add an annotation to the bars you’ll have to use either
geom_label(). I will start off with the former. Both require the
label aesthetic which tells
ggplot2 which text to actually display. In addition, both functions require the
y aesthetics but these are already set when using
bar_chart() so I won’t bother setting them explicitly after this first example.
chart + geom_text(aes(x = language, y = pct, label = pct))
By default the labels are center-aligned directly at the
y value. You will never want to leave it like that because it’s quite hard to read. To left-align the labels set the
hjust parameter to
chart + geom_text(aes(label = pct, hjust = "left"))
That’s still not ideal I would say. Let’s move the labels a bit further away from the bars by setting
hjust to a negative number and increase the axis limits to improve the legibility of the label of the top most bar.
chart + geom_text(aes(label = pct, hjust = -0.2)) + ylim(NA, 100)
Alternatively, you may want to have the labels inside the bars.
chart + geom_text(aes(label = pct, hjust = 1))
Again, a bit close to the end of the bars. By increasing the
hjust value the labels can be moved further to the left. In addition, black on blue is quite hard to read so let’s change the text color to white. Notice that this happens outside of
chart + geom_text(aes(label = pct, hjust = 1.2), color = "white")
Next, let’s try
geom_label() for once to see how it’s different from
chart + geom_label(aes(label = pct, hjust = 1.2))
I am not a fan of this look and will stick to
geom_text() for the final plot.
As the data in the plot represents percentages it’s best practice to have the labels include the percentage sign. In addition, let’s highlight our favorite language
R and add title, footnotes etc.
dreaded_lang %>% mutate(label = sprintf("%1.1f%%", pct)) %>% bar_chart(language, pct, highlight = "R", bar_color = "black") + geom_text(aes(label = label, hjust = -0.1), size = 5) + scale_y_continuous( limits = c(0, 100), expand = expansion() ) + labs( x = NULL, y = "Developers Who are Developing with the Language but
Have not Expressed Interest in Continuing to Do so", title = "Top 10 Most Dreaded Programming Languages", subtitle = "*R Placed 8th*", caption = "Source: Stackoverflow Developer Survey 2019" ) + mdthemes::md_theme_classic(base_size = 14) + theme( axis.text.x = element_blank(), axis.line.x = element_blank(), axis.ticks.x = element_blank() )
Notice how easy it was to highlight a single bar thanks to
ggcharts. In addition, I used my
mdthemes package which provides themes that interpret text as markdown. That way is was super easy to get the subtitle in italics. Furthermore, I removed the axis labels and grid lines. In my opinion you should never have an axis and labels in the same plot.
To finish off this post, let’s have a quick look at how to label a vertical bar chart. It’s basically the same process but instead of using
hjust you will need to use
vjust to adjust the label position.
data("biomedicalrevenue") biomedicalrevenue %>% filter(company == "Novartis") %>% mutate(label = sprintf("$%1.2f B.", revenue)) %>% column_chart(year, revenue) + geom_text(aes(label = label, vjust = -1)) + theme( axis.text.y = element_blank(), panel.grid.major.y = element_blank() ) + scale_x_continuous( name = "Fiscal Year", breaks = 2011:2018 ) + scale_y_continuous( name = "Revenue (Billion USD)", limits = c(0, 70), expand = expansion() )