Boxenplot (letter-value plot) in seaborn

Sample data

Copy the following code to get the data that will be used in the following examples.

import numpy as np
import seaborn as sns
import random

# Data simulation
rng = np.random.RandomState(5)
variable = rng.normal(0, 1, size = 2000)
random.seed(5)
group = random.choices(["G1", "G2", "G3"], k = 2000)
df = {'variable': variable, 'group': group}

Boxenplot in seaborn with boxenplot

A boxenplot (also called a letter-value plot) is a box plot with more quantile levels, which shows more detail about the shape of the distribution in the tails. It is most useful for larger data sets, where a regular box plot only shows the outliers as individual points instead of extra boxes.

import seaborn as sns

sns.boxenplot(x = "variable", data = df)

# Equivalent to:
sns.boxenplot(x = variable)

Grouped boxenplot

As with a regular box plot, pass a categorical variable to y (or x for a horizontal orientation) to draw one boxenplot per group.

import seaborn as sns

sns.boxenplot(x = "group", y = "variable", data = df)

Colors

Use palette to assign a color per group, or color for a single fixed color.

import seaborn as sns

sns.boxenplot(x = "group", y = "variable", data = df, palette = "deep")

Number of levels

The number of quantile boxes is controlled by k_depth, which defaults to "tukey". Pass an integer to fix the number of levels manually, or "proportion" / "trustworthy" to let seaborn pick it based on other rules.

import seaborn as sns

sns.boxenplot(x = "variable", data = df, k_depth = 3)

Outline width

The width and color of the box outlines can be set with linewidth and linecolor.

import seaborn as sns

sns.boxenplot(x = "group", y = "variable", data = df,
              linewidth = 1.5, linecolor = "black")