Violin plot by group in ggplot2

Sample data

The data set used in the following examples is warpbreaks, which gives the number of warp breaks in a yarn, the type of wool and the level of tension.

# Sample data set
warpbreaks

Grouped violin plot with `geom_violin`

A violin plot by group can be created in ggplot passing the numerical (breaks) and the categorical (tension) variable to aes and using geom_violin.

# install.packages("ggplot2")
library(ggplot2)

ggplot(warpbreaks, aes(x = tension, y = breaks)) +
  geom_violin()

Violin plot by group in ggplot2

Horizontal violin plot

If you want a horizontal violin plot instead of vertical you can pass the categorical variable (tension) to y or use coord_flip as in the example below.

# install.packages("ggplot2")
library(ggplot2)

ggplot(warpbreaks, aes(x = tension, y = breaks)) +
  geom_violin() +
  coord_flip()

Horizontal violin plot by group in ggplot2

Avoid trimming the trails

By default, the trails of the violin are trimmed to the range of the data. To avoid trimming set trim = FALSE inside geom_violin.

# install.packages("ggplot2")
library(ggplot2)

ggplot(warpbreaks, aes(x = tension, y = breaks)) +
  geom_violin(trim = FALSE)

Violin plot ggplot2 without trimmed trails

Adding quantiles

The desired quantiles can be added passing a vector to the draw_quantiles argument, as in the example below.

# install.packages("ggplot2")
library(ggplot2)

ggplot(warpbreaks, aes(x = tension, y = breaks)) +
  geom_violin(trim = FALSE,
              draw_quantiles = c(0.25, 0.5, 0.75))

Add quantiles and quartiles to ggplot2 violin plot

Adding box plots

You can also overlay box plots to the violin plots to show the median and the outliers. Recall to set a small width for the box plots.

# install.packages("ggplot2")
library(ggplot2)

ggplot(warpbreaks, aes(x = tension, y = breaks)) +
  geom_violin(trim = FALSE) +
  geom_boxplot(width = 0.07)

Violin plot with box plot in ggplot2

Bandwidth customization

The violins are kernel density estimates of the data, so a bandwidth is involved for its computation. The default bandwidth can be changed with bw.

# install.packages("ggplot2")
library(ggplot2)

ggplot(warpbreaks, aes(x = tension, y = breaks)) +
  geom_violin(trim = FALSE, bw = 10) +
  geom_boxplot(width = 0.07)

Bandwidth selection violin plot in R

Fill and border colors

Violin plot fill color by group

Fill color by group

If you want to fill the violins by group pass the categorical variable to the fill argument of aes.

# install.packages("ggplot2")
library(ggplot2)

ggplot(warpbreaks, aes(x = tension, y = breaks, fill = tension)) +
  geom_violin(trim = FALSE) +
  geom_boxplot(width = 0.07)

Grouped violin plot in ggplot2

Fill color by subgroup

If you have other categorical variable you can create subgroups and fill the areas based on these subgroups.

# install.packages("ggplot2")
library(ggplot2)

ggplot(warpbreaks, aes(x = tension, y = breaks, fill = wool)) +
  geom_violin(trim = FALSE) +
  geom_boxplot(width = 0.07, position = position_dodge(width = 0.9))

Violin plot in ggplot2 with brewer palette

Color scale

The default colors can be changed. For instance, you can use the brewer palette as follows:

# install.packages("ggplot2")
library(ggplot2)

ggplot(warpbreaks, aes(x = tension, y = breaks, fill = tension)) +
  geom_violin(trim = FALSE) + +
  geom_boxplot(width = 0.07)
  scale_fill_brewer()

Violin plot color customization in R

Custom colors

If you want to use your custom color palette you can use scale_fill_manual and input the colors to the values argument.

# install.packages("ggplot2")
library(ggplot2)

ggplot(warpbreaks, aes(x = tension, y = breaks, fill = tension)) +
  geom_violin(trim = FALSE) + 
  geom_boxplot(width = 0.07) +
  scale_fill_manual(values = c("#BCE4D8", "#49A4B9", "#2C5985"))

Violin plot fill transparency with geom_violin

Fill transparency

The fill transparency can be modified with the alpha argument of the geom_violin function.

# install.packages("ggplot2")
library(ggplot2)

ggplot(warpbreaks, aes(x = tension, y = breaks, fill = tension)) +
  geom_violin(trim = FALSE,
              alpha = 0.5) +
  geom_boxplot(width = 0.07)

geom_violin border color

Border color

The color argument of geom_violin can be used to change the color of the borders.

# install.packages("ggplot2")
library(ggplot2)

ggplot(warpbreaks, aes(x = tension, y = breaks, fill = tension)) +
  geom_violin(trim = FALSE,
              color = "blue") +
  geom_boxplot(width = 0.07)

Violin border color by group in R

Border color by group

However, if you want to set a border color based on the groups, you can pass the categorical variable to the color argument of the aes function.

# install.packages("ggplot2")
library(ggplot2)

ggplot(warpbreaks, aes(x = tension, y = breaks, color = tension)) +
  geom_violin(trim = FALSE) +
  geom_boxplot(width = 0.07)

Custom border color ggplot violin plots

Custom border colors

Similarly to changing the fill colors, you can customize the border colors, but with scale_color_manual.

# install.packages("ggplot2")
library(ggplot2)

ggplot(warpbreaks, aes(x = tension, y = breaks, color = tension)) +
  geom_violin(trim = FALSE) + 
  geom_boxplot(width = 0.07) +
  scale_color_manual(values = c("#F4D166", "#EC6E1C", "#B71D3E"))

Legend customization

Legend title

The legend title displays the name of the categorical variable. To change this default title use the guides function as follows.

# install.packages("ggplot2")
library(ggplot2)

ggplot(warpbreaks, aes(x = tension, y = breaks, fill = tension)) +
  geom_violin(trim = FALSE) + 
  geom_boxplot(width = 0.07) +
  guides(fill = guide_legend(title = "Title"))

Change the legend title ggplot2 violin plot

Key labels

The legend key labels are the names of the groups. These labels can be changed with scale_fill_manual if you change the fill colors or with scale_fill_hue to only change the labels.

# install.packages("ggplot2")
library(ggplot2)

ggplot(warpbreaks, aes(x = tension, y = breaks, fill = tension)) +
  geom_violin(trim = FALSE) + 
  geom_boxplot(width = 0.07) +
  scale_fill_hue(labels = c("G1", "G2", "G3"))

Change the legend labels ggplot2 violin plots

Remove legend

Finally, if you want to remove the default legend you can set the legend position to "none" or add show.legend = FALSE to geom_violin and geom_boxplot (if you added it).

# install.packages("ggplot2")
library(ggplot2)

ggplot(warpbreaks, aes(x = tension, y = breaks, fill = tension)) +
  geom_violin(trim = FALSE) + 
  geom_boxplot(width = 0.07) + 
  theme(legend.position = "none")

Remove the legend from a ggplot2 violin plot