Seaborn

Seaborn is a high-level interface to matplotlib. It is great for rapidly creating standard yet pretty data visualizations.

Pros:

• Easy-to-use, high-level
• Pretty statistical data visualizations by default
• Theming and color palettes
• Extendable or modifable with matplotlib code

Cons:

• Limited support for complex visualizations (in which case, consider matplotlib)

To install, run:

pip install seaborn

Details For use in Jupyter notebooks, make sure you enable matplotlib mode to see the plots. Alternatively, call:

import matplotlib.pyplot as plt
plt.show()

Code

import seaborn as sns
import seaborn.objects as so

import matplotlib as mpl
import matplotlib.pyplot as plt

import numpy as np
import pandas as pd

sns.set_theme()

Theming

If no theme is set, the default matplotlib theme is used.

Action	Code	Details
Set default theme	sns.set_theme()

Plots

Distribution plots

See https://seaborn.pydata.org/generated/seaborn.displot.html

Boxplots

Action	Code	Details
Boxplot	sns.boxplot(iris, x='sepal_length')
Boxplot by group	sns.boxplot(iris, x='sepal_length', y='species')
Facetted boxplot by two grouping factors	sns.catplot(fmri, kind='box', x='signal', y='region', col='event')
Boxplot by two grouping factors	sns.boxplot(fmri, x='signal', y='region', hue='event')

Histograms

Action	Code	Details
Plot histogram	sns.histplot(iris, x='sepal_length')
Plot histogram with bins centered at discrete values	sns.histplot(iris, x='sepal_length', discrete=True)	For example, at integer values
Plot histogram with bin-width	sns.histplot(iris, x='sepal_length', binwidth=1.0)
Plot histogram with a given number of bins	sns.histplot(iris, x='sepal_length', bins=5)
Plot histogram and kernel density	sns.histplot(iris, x='sepal_length', kde=True)
Plot normalized histogram	sns.histplot(iris, x='sepal_length', stat='proportion')
Plot normalized histogram as percentage	sns.histplot(iris, x='sepal_length', stat='percent')
Facetted histogram chart	sns.displot(iris, x='sepal_length', col='species')

Kernel density plots

Action	Code	Details
Plot kernel density	sns.kdeplot(iris, x='sepal_length')
Plot kernel density by group	sns.kdeplot(iris, x='sepal_length', hue='species')
Facetted kernel density chart	sns.displot(iris, kind='kde', x='sepal_length', col='species')
Violin plot	sns.violinplot(iris, x='sepal_length')
Violin plot with grouping factor	sns.violinplot(iris, x='sepal_length', y='species')

Empirical cumulative density plots

Action	Code	Details
Plot stepped cumulative density	sns.ecdfplot(iris, x='sepal_length')
Plot stepped inverse cumulative density	sns.ecdfplot(iris, x='sepal_length', complementary=True)
Plot stepped cumulative density by group	sns.displot(iris, kind='ecdf', x='sepal_length', hue='species')
Facetted plot of stepped cumulative density	sns.displot(iris, kind='ecdf', x='sepal_length', col='species')
Plot cumulative density as barplot	sns.histplot(iris, x='sepal_length', cumulative=True, stat='proportion')
Plot cumulative density by group as barplot	sns.histplot(iris, x='sepal_length', hue='species', cumulative=True, stat='proportion', common_norm=False)	Each group reaches 1.0
Facetted plot of cumulative density by group as barplot	sns.displot(iris, x='sepal_length', cumulative=True, stat='proportion', common_norm=False, col='species')	Each group reaches 1.0
Plot step-wise cumulative density with density rug	sns.displot(iris, kind='ecdf', x='sepal_length', rug=True)

Bivariate plots

Plots involving two continuous variables.

Scatter plots

Action	Code	Details
Scatter plot	sns.scatterplot(iris, x='sepal_length', y='sepal_width')
Scatter plot with grouping factor	sns.scatterplot(iris, x='sepal_length', y='sepal_width', hue='species')
Facetted scatter plot	sns.relplot( data=iris, kind='scatter', x='sepal_length', y='sepal_width', col='species' )

Joint plots

Action	Code	Details
Scatter plot with KDE axes	sns.jointplot(iris, x='sepal_length', y='sepal_width')	Scatter and KDE
Scatter plot with KDE axes, with grouping factor	sns.jointplot(iris, x='sepal_length', y='sepal_width', hue='species')
Plot heatmap with histograms along axes	sns.jointplot(iris, kind='hist', x='sepal_length', y='sepal_width')

Line plots

Action	Code	Details
Line plot	sns.lineplot(flights.query('month=="May"'), x='year', y='passengers', hue='month')
Line plot, with separate lines per group	sns.lineplot(flights, x='year', y='passengers', hue='month')
Line plot involving duplicate observations per x, as confidence region	sns.lineplot(flights, x='year', y='passengers')
Facetted line plot	sns.relplot( data=fmri, kind='line', x='timepoint', y='signal', col='region', row='event' )

Multivariate plots

Plots involving more than two continuous variables.

Action	Code	Details
Pairs plot	sns.pairplot(iris)	Plots pairwise scatter, and KDE along the diagonal

Facetted plotting options

The function for creating facetted plots differs between the kinds of plots, so only options are described here. See, for example, sns.displot(), sns.relplot() and sns.catplot().

Action	Code	Details
Horizontally stacked facets	col='region'
Horizontally stacked facets, wrap after n columns	col='region', col_wrap=n
Vertically stacked facets	row='region'
Vertically stacked facets, wrap after n columns	row='region', row_wrap=n
Grid-based facets along two grouping factors	col='region', row='event'

Plot configuration

Arguments should be used inside the sns.relplot function.

Action	Code	Details
Flip axes	orient='y'
Log-scale with base b	log_scale=('y', b)
Format axis with comma as thousands separator	import matplotlib.ticker as ticker iris['sepal_length'] *= 1000 g = sns.histplot(iris, x='sepal_length') g.xaxis.set_major_formatter( ticker.FuncFormatter(lambda x, p: format(int(x), ',')) )	Very tedious
Format axis with percentage labels	import matplotlib.ticker as ticker g = sns.ecdfplot(iris, x='sepal_length', stat='percent') g.yaxis.set_major_formatter( ticker.PercentFormatter() )	PercentFormatter assumes percent input, not proportion, so 1.0 = '1.0%'
1:1 aspect ratio	aspect=1
Hide legend	legend=False