Review gghighlight - a great package to highlight ggplot charts

I have been watching gghighlight for couple of months as his package addresses one of problems that I very much care about: “Highlighting in ggplot”. Recently, the package has released in CRAN with big improvement since the first version. So, I decided to test this package in some common charts in ggplot2

First, let’s set up some theme to use during the experiment.

#Set up library & theme

library(dplyr)
library(ggplot2)
library(gghighlight)

my_theme <- function(){
  list(
      theme_bw(),
      scale_fill_brewer(palette = "Set1"),
      scale_color_brewer(palette = "Set1")
    )
}

Highlight bar chart

data <- mtcars %>% mutate(cyl = factor(cyl))

data %>% 
  ggplot(aes(cyl, mpg)) +
  geom_bar(aes(fill = cyl), stat = "identity") +
  my_theme() +
  gghighlight(cyl == 6) +
  labs(title = "Example 1 - highlight without text") 

  • Highlight bar chart with text
data %>% 
  group_by(cyl) %>% 
  summarise(mpg = sum(mpg)) %>% 
  ggplot(aes(cyl, mpg)) +
  geom_bar(aes(fill = cyl), stat = "identity") +
  geom_text(aes(label = mpg),
                       hjust = -0.3) +
  my_theme() +
  coord_flip() +
  scale_y_continuous(limits = c(0, 320)) +
  gghighlight(cyl == 6) +
  labs(title = "Example 2 - Hightlight with text (version 1)")

As you can see, text in columns where cyl equal to 4 & 8 still appears, which is not what I expected. However, by moving geom_text layer after highlighting layer, I can hide all texts which I don’t want to highlight. Let’s check another example:

data %>% 
  group_by(cyl) %>% 
  summarise(mpg = sum(mpg)) %>% 
  ungroup %>% 
  ggplot(aes(cyl, mpg)) +
  geom_bar(aes(fill = cyl), stat = "identity") +
  my_theme() +
  coord_flip() +
  scale_y_continuous(limits = c(0, 320)) +
  gghighlight(cyl == 6) +
  geom_text(aes(label = mpg),
                       hjust = -0.3) +
    labs(title = "Example 3 - Hightlight with text (version 2)")

Boom! It created the plot I desired exactly! This functionality is so great that I shared with my coleagues right after realizing how cool it is & all members of my team were very impressed!

Let’s check other use cases where gghighlight could change the way we emphasize readers to get insights from charts.

Boxplot

iris %>% 
  ggplot(aes(Species, Sepal.Length)) +
  geom_boxplot(aes(fill = Species)) +
  my_theme() +
  facet_wrap(~Species) +
  gghighlight() +
  theme(legend.position = "top") +
  labs(title = "Example 4 - highlight for box plot")

Scatter plot

With gghighlight, it is very easy to highlight interesting points in a graph. You can add different layers after gghighlight to emphasize your poitns. For example, I want to create a scatter plot in mtcars with mpg & disp being horizontal & vertical axis respectively and emphasize some points under following conditions:

  • Emphasize points with disp between 350 & 400
  • Add names of cars for these above points
  • Make emphasized points in different color & bigger

The solution utilizing gghighlight is so simple as follows:

df <- mtcars %>% mutate(name = row.names(.))
df %>% 
  ggplot(aes(mpg, disp)) +
  geom_point(col = "darkred") +
  my_theme() +
  gghighlight(disp > 350 & disp <= 400,
              unhighlighted_colour = alpha("darkgreen", 0.4),
              use_direct_label = T,
              label_key = name,
              label_params = list(size = 5)) +
  geom_point(col = "darkred", size = 2.5) +
  labs(title = "Example 5 - Hightlight points")

After practicing some examples, I do fall in love with this greate R package! It will add much value to ggplot2 it has been already added in my daily packges to use in R! Highly recommended package for all R users and many thanks Hiroaki Yutani for creating such a great R package!

Anh Hoang Duc
Anh Hoang Duc
Business Analytics professional

.

comments powered by Disqus

Related