Word cloud in ggplot2 with ggwordcloud

Sample data

# install.packages("ggwordcloud")
library(ggwordcloud)

df <- thankyou_words_small

Word cloud with ggwordcloud

Basic word cloud

ggwordcloud provides a ggplot2 geom named geom_text_wordcloud for creating word clouds. Use your data frame and pass the column containing the texts to the label argument of aes and use the geom_text_wordcloud function. Note that we are setting a seed to keep the example reproducible, as the algorithm used for placing the texts involves some randomness.

# install.packages("ggwordcloud")
library(ggwordcloud)

# Data
df <- thankyou_words_small

set.seed(1)
ggplot(df, aes(label = word)) +
  geom_text_wordcloud() +
  theme_minimal()

Size of the text based on a variable

So far all the words had the same size. If you want to set the size based on a numerical variable you can pass it to the size argument of aes.

# install.packages("ggwordcloud")
library(ggwordcloud)

# Data
df <- thankyou_words_small

set.seed(1)
ggplot(df, aes(label = word, size = speakers)) +
  geom_text_wordcloud() +
  theme_minimal()

Basic word cloud with base R syntax

Alternatively, you could use the ggwordcloud and specify the words and frequency (which will determine the relative size of each text) to create the word cloud with a single function. Note that this function provides more arguments which you can customize.

# install.packages("ggwordcloud")
library(ggwordcloud)

# Data
df <- thankyou_words_small

set.seed(1)
ggwordcloud(words = df$word, freq = df$speakers)

Scaling (font size)

The default text scaling of ggplot2 (square root scaling) makes the word cloud look small respect to the plot area. For this reason, you could use the scale_size_area function as follows to obtain a better font size control.

# install.packages("ggwordcloud")
library(ggwordcloud)

# Data
df <- thankyou_words_small

set.seed(1)
ggplot(df, aes(label = word, size = speakers)) +
  geom_text_wordcloud() +
  scale_size_area(max_size = 20) +
  theme_minimal()

Removing texts that not fit

If you have too many words and a big font size you can set the rm_outside argument of geom_text_wordcloud to TRUE or decrease the font size to remove the overflowing texts.

# install.packages("ggwordcloud")
library(ggwordcloud)
# install.packages("ggforce")
library(ggforce)

# Data
df <- thankyou_words_small

set.seed(1)
ggplot(df, aes(label = word, size = speakers)) +
  geom_text_wordcloud(rm_outside = TRUE) +
  scale_size_area(max_size = 60) +
  theme_minimal()

Text rotation

Note that you can also rotate the texts with the angle argument of aes. In the following example we are creating a new column randomly to represent the desired angles to rotate each text.

# install.packages("ggwordcloud")
library(ggwordcloud)

set.seed(1)

# Data
df <- thankyou_words_small
df$angle <- sample(c(0, 45, 60, 90, 120, 180), nrow(df), replace = TRUE)

ggplot(df, aes(label = word, size = speakers, angle = angle)) +
  geom_text_wordcloud() +
  scale_size_area(max_size = 20) +
  theme_minimal()

Shape of the word cloud

By default, the shape of the word cloud is circular. However, it is possible to change the shape of the cloud with the shape argument of the geom_text_wordcloud function. Possible shapes are named "circle" (default), "cardioid", "diamond", "pentagon", "star", "square", "triangle-forward" and "triangle-upright". In the following blocks of code you can check a couple of examples.

# install.packages("ggwordcloud")
library(ggwordcloud)

# Data
df <- thankyou_words_small

set.seed(1)
ggplot(df, aes(label = word, size = speakers)) +
  geom_text_wordcloud(shape = "diamond") +
  scale_size_area(max_size = 20) +
  theme_minimal()

# install.packages("ggwordcloud")
library(ggwordcloud)

# Data
df <- thankyou_words_small

set.seed(1)
ggplot(df, aes(label = word, size = speakers)) +
  geom_text_wordcloud(shape = "star") +
  scale_size_area(max_size = 20) +
  theme_minimal()

Using a mask

Alternatively you can use a PNG file to create a mask to place the words within it. Note that the non-transparent pixels of the image will be used as the mask. In the following example we are using a sample PNG file from the ggwordcloud package with the shape of a heart to create the mask.

# install.packages("ggwordcloud")
library(ggwordcloud)

# Data
df <- thankyou_words_small

# Mask
mask_png <- png::readPNG(system.file("extdata/hearth.png",
      package = "ggwordcloud", mustWork = TRUE))

set.seed(1)
ggplot(df, aes(label = word, size = speakers)) +
  geom_text_wordcloud(mask = mask_png) +
  scale_size_area(max_size = 20) +
  theme_minimal()

Color of the texts

# install.packages("ggwordcloud")
library(ggwordcloud)

# Data
df <- thankyou_words_small

set.seed(1)
ggplot(df, aes(label = word, size = speakers)) +
  geom_text_wordcloud(color = "red") +
  scale_size_area(max_size = 20) +
  scale_color_discrete("red") +
  theme_minimal()

# install.packages("ggwordcloud")
library(ggwordcloud)

# Data
df <- thankyou_words_small

set.seed(1)
ggplot(df, aes(label = word, size = speakers, color = name)) +
  geom_text_wordcloud() +
  scale_size_area(max_size = 20) +
  theme_minimal()

# install.packages("ggwordcloud")
library(ggwordcloud)

# Data
df <- thankyou_words_small

set.seed(1)
ggplot(df, aes(label = word, size = speakers, color = speakers)) +
  geom_text_wordcloud() +
  scale_size_area(max_size = 20) +
  theme_minimal() +
  scale_color_gradient(low = "darkred", high = "red")