Align violin plots with dodged box plots

14,106

You need to set the width of the dodging explicitly for both geoms:

dodge <- position_dodge(width = 0.4)

ggplot(data = df, aes(x = groups, y = value, fill = type)) +
  geom_violin(position = dodge)+
  geom_boxplot(width=.1, outlier.colour=NA, position = dodge) 

enter image description here


For a more thorough explanation, see What is the width argument in position_dodge?

Share:
14,106

Related videos on Youtube

Sergio.pv
Author by

Sergio.pv

Updated on May 20, 2022

Comments

  • Sergio.pv
    Sergio.pv about 2 years

    I have this data frame

    set.seed(1234)
    x <- rnorm(80, 5, 1)
    df <- data.frame(groups = c(rep("group1",20),
                               rep("group2",20),
                               rep("group3",20),
                               rep("group4",20)),
                        value = x,
                        type = c(rep("A", 10),
                               rep("B", 10),
                               rep("A", 10),
                               rep("B", 10),
                               rep("A", 10),
                               rep("B", 10),
                               rep("A", 10),
                               rep("B", 10)))
    

    And I would like to plot it as violin plot, aligned with a narrow box plot and grouped by 'type':

    ggplot(data = df, aes(x = groups, y = value, fill = type)) +
      geom_violin()+
      geom_boxplot(width = 0.1, outlier.colour = NA)
    

    However, the box plots does not align to the violin plots. What is the missing argument to tell ggplot to do such overlying?

    Thanks!

    • Roland
      Roland over 9 years
      I'm not sure what you mean exactly by "align". Possibly set the position? See help("position_dodge").
  • svavil
    svavil almost 4 years
    I'm looking for a way to make boxplots filled with white, while simultaneously retaining positions of both boxplots and violins, and keeping violins colored. Do you know a way to reach this?
  • Henrik
    Henrik almost 4 years
    @svavil Yes, in geom_boxplot, use the group argument instead of fill. ggplot(data = df, aes(x = groups, y = value)) + geom_violin(aes(fill = type), position = dodge) + geom_boxplot(aes(group = interaction(type, groups)), width=.1, outlier.colour=NA, position = dodge). See ?aes_group_order.