seaborn’s statistical charts have direct counterparts here, with the same vocabulary for what a group is reduced to and how its uncertainty is shown. As with every chart in dataviz, each function returns an unrendered chart that you export with save().

Charts

seaborn	dataviz
scatterplot(x=, y=)	scatter(x, y)
scatterplot(..., hue=g)	Plot().mark_point().encode(x=x, y=y, color_categories=g)
scatterplot(..., size=s)	encode(..., size=s)
stripplot(..., jitter=True)	Plot().mark_point(jitter_x=..., jitter_y=...)
swarmplot	beeswarm
lineplot (with repeated x)	lineplot – the per-x estimate with a band
barplot	barplot
pointplot	pointplot
boxplot / violinplot / boxenplot	box / violin / boxenplot
histplot	histogram
kdeplot / rugplot / ecdfplot	kdeplot / rugplot / ecdf
heatmap / clustermap	heatmap / clustermap
pairplot / jointplot	pairplot / jointplot
residplot	residplot
regplot	scatter(x, y).annotate_best_fit() – the line and its confidence band
FacetGrid	render_facets(a, b, cols=...), one chart per facet
despine()	nothing to do: the top and right axis lines are off by default

Estimates and error bars

barplot, pointplot and lineplot reduce each group with an Estimator and size its uncertainty with an ErrorBar, as seaborn’s estimator= and errorbar= do:

seaborn	dataviz
estimator="mean" (default) / "median" / len / sum	Estimator.MEAN / MEDIAN / COUNT / SUM
errorbar=("ci", 95) (default)	ErrorBar.ci(0.95) – a bootstrap confidence interval with a fixed seed
errorbar=("pi", 50)	ErrorBar.pi(0.5) – a percentile interval of the observations
errorbar="se" / ("se", 2)	ErrorBar.se() / ErrorBar.se(2.0)
errorbar="sd"	ErrorBar.sd()
errorbar=None	ErrorBar.none()

Levels are fractions here (0.95), not percentages.

Behavior that differs on purpose

• The estimator names the value axis. When no axis title is given, barplot() and pointplot() label the value axis “Mean”, “Median”, “Count” or “Sum”, so an estimate is never mistaken for raw data.
• Bootstrap intervals are reproducible. ErrorBar.ci() resamples with a fixed seed (the seed= argument), so the same data renders the same whisker every time.
• A pair plot’s diagonal is a true histogram on the same numeric axis as the scatter panels in its column, so the bars line up with the points above and below them.