Skip to content

22 - Seaborn

Previous: 21 - Matplotlib | Index: All guides | Next: 23 - Scikit-learn

Quick reference for statistical plots with Seaborn (built on matplotlib, works directly with pandas DataFrames).

Last verified: 2026-09-27. For newer changes, check the Official docs links in the Introduction.

Introduction

Before you start

You should know: pandas DataFrames (18) and the Figure / Axes idea from Matplotlib (21), because Seaborn draws with Matplotlib underneath.

The problem it solves: common statistical charts ("distribution of prices per category, split by region, with averages and error bars") take many lines of Matplotlib: grouping the data, choosing colours, adding legends. Most of that is the same every time.

Before Seaborn: you wrote that grouping and styling code by hand for every chart, or switched to R's ggplot2 for statistical graphics. Seaborn (2012) brought that "describe the data, get the chart" style to Python, on top of Matplotlib.

Think of it like: chart templates with good taste built in. You name the columns and the kind of chart; Seaborn does the grouping, colours and legend, and you can still adjust details with Matplotlib afterwards.

What is Seaborn?

Seaborn is a statistical visualisation library built on Matplotlib. You pass a pandas DataFrame and column names, and it creates attractive, informative charts in one line: distributions, comparisons between categories, relationships, regression lines and correlation heatmaps. It automatically handles grouping (hue), colours, legends and confidence intervals.

Why use it?

  • Less code: one call replaces many lines of Matplotlib.
  • Built for DataFrames: refer to columns by name.
  • Statistics included: averages with error bars, density curves, fitted regression lines.
  • Good defaults: nice themes and colour palettes out of the box.
  • Fast exploration: pairplot or heatmap of correlations gives a quick overview of a dataset.

Key terms

Term Meaning
hue Column used to colour groups
Axes-level function Draws one plot into a Matplotlib Axes (boxplot)
Figure-level function Creates a whole figure, can split into facets (catplot)
Facet One small plot per category value (col=, row=)
Palette Set of colours
KDE Smooth estimate of a distribution

Where it fits: needs 18 - Pandas data; customise with 21 - Matplotlib.

Official docs

Where to read the latest, authoritative documentation:

Resource Link
seaborn documentation https://seaborn.pydata.org/
seaborn API reference https://seaborn.pydata.org/api.html
seaborn example gallery https://seaborn.pydata.org/examples/index.html

Contents

  1. Flags and Parameters
  2. Install and Import
  3. Built-in Datasets
  4. How Seaborn Works
  5. Theme and Style
  6. Distribution Plots
  7. Categorical Plots
  8. Relationship Plots
  9. Regression Plots
  10. Heatmap and Correlation
  11. Pair Plot and Joint Plot
  12. Figure-level vs Axes-level
  13. Facets (Grid of Plots)
  14. Colors and Palettes
  15. Customize with Matplotlib
  16. Save a Figure
  17. Which Plot to Use
  18. Troubleshooting
  19. Try It

0. Flags and Parameters

The meaning of the arguments and parameters used in the seaborn calls below. Explains how a call is built, then lists each parameter with its meaning and example.

Use this when you see sns.catplot(data=tips, x="day", y="tip", kind="box", col="time") and want to know what each argument does.

How a function call is built

sns.boxplot(data=tips, x="day", y="tip", hue="sex")
|   |       |          |        |        |
|   |       |          |        |        +-- colour by this column
|   |       |          |        +----------- column for the y axis
|   |       |          +-------------------- column for the x axis
|   |       +------------------------------- the DataFrame to use
|   +--------------------------------------- plot type
+------------------------------------------- seaborn module
  • Positional arguments come first, in a fixed order. Keyword arguments use name=value and can be in any order.
  • Arguments you leave out use their default value (for example no hue = one colour).
  • See all parameters and defaults: help(sns.boxplot), or Shift+Tab inside the brackets in Jupyter.
Parameter Used in Meaning Example
data all The DataFrame data=tips
x, y all Column names for the axes x="day", y="tip"
hue most Colour by this column hue="sex"
size scatterplot, relplot Marker size by this column size="size"
style scatterplot, lineplot Marker / line style by this column style="time"
col, row figure-level (relplot, catplot, displot, lmplot) One subplot per value of this column col="time"
col_wrap figure-level Max subplots per row col_wrap=2
kind catplot, relplot, displot, jointplot Which plot to draw kind="box"
ax axes-level Draw into this matplotlib Axes ax=axes[0]
height, aspect figure-level Height of each subplot (inches), width = height x aspect height=4, aspect=1.5
palette most Colour set for hue palette="Set2"
color most One colour for everything color="steelblue"
order, hue_order categorical Order of categories order=["Thur", "Fri"]
legend most False = hide the legend legend=False
bins histplot Number of bars bins=20
kde histplot, displot Add a smooth density curve kde=True
multiple histplot Several hue groups: "layer", "stack", "dodge", "fill" multiple="stack"
fill kdeplot Fill the area under the curve fill=True
estimator barplot, pointplot What the bar height shows: "mean" (default), "sum", "median" estimator="sum"
errorbar barplot, lineplot Error bar type: ("ci", 95) default, "sd", None errorbar=None
split violinplot Two hue groups as halves of one violin split=True
jitter stripplot Spread points sideways so they overlap less jitter=True
ci regplot, lmplot Confidence band around the line; None = hide ci=None
order regplot Polynomial degree of the fitted line (not category order here) order=2
scatter_kws regplot Extra options for the points, as a dict scatter_kws={"alpha": 0.5}
annot heatmap Write the value in each cell annot=True
fmt heatmap Number format of annot fmt=".2f"
cmap heatmap Colour map cmap="coolwarm"
vmin, vmax, center heatmap Colour scale limits and middle value vmin=-1, vmax=1, center=0
linewidths heatmap Gap between cells linewidths=0.5
corner pairplot Only lower triangle (no duplicates) corner=True
diag_kind pairplot Plot on the diagonal: "hist" or "kde" diag_kind="kde"
vars pairplot Only these columns vars=["tip", "size"]
style, context, font_scale set_theme Background style, overall size, font size factor style="whitegrid"

1. Install and Import

Installing and importing seaborn. pip install seaborn, then import seaborn as sns (plus matplotlib to show / save).

Use it for statistical charts from DataFrames with less code than matplotlib.

pip install seaborn

import seaborn as sns
import matplotlib.pyplot as plt
import pandas as pd

2. Built-in Datasets

Sample datasets that ship with seaborn. sns.load_dataset(name) downloads a small DataFrame.

Use it for practising, testing a plot type, or reproducing documentation examples.

sns.get_dataset_names()             # list available datasets (needs internet)
tips = sns.load_dataset("tips")     # restaurant bills and tips
iris = sns.load_dataset("iris")     # flower measurements
titanic = sns.load_dataset("titanic")

Examples below use tips (columns: total_bill, tip, sex, smoker, day, time, size).

3. How Seaborn Works

The common pattern of every seaborn function. Pass the DataFrame as data= and column names to x, y, hue. Read once; every function in this guide works this way.

Pass a DataFrame to data= and column names to x=, y=, hue=:

sns.scatterplot(data=tips, x="total_bill", y="tip", hue="time")
plt.show()
Parameter Meaning
data DataFrame
x, y column names
hue color by this column
size marker size by this column
style marker / line style by this column
col, row split into subplots (figure-level functions only)
palette color set
order, hue_order category order
ax draw into an existing matplotlib Axes

4. Theme and Style

Global look: background, grid, font size, colours. sns.set_theme(style=..., context=..., palette=...) affects all following plots.

Use it at the start of a notebook to make all charts consistent and presentation-ready.

sns.set_theme()                                     # default seaborn look
sns.set_theme(style="whitegrid", palette="deep", font_scale=1.2)
sns.set_style("ticks")                              # darkgrid, whitegrid, dark, white, ticks
sns.set_context("talk")                             # paper, notebook, talk, poster (size)
sns.despine()                                       # remove top and right borders

5. Distribution Plots

How values of one variable are spread. histplot, kdeplot, ecdfplot; hue compares groups.

Use it for checking skew, outliers and differences between groups before modelling.

One variable: how are values spread?

sns.histplot(data=tips, x="total_bill")                     # histogram
sns.histplot(data=tips, x="total_bill", bins=20, kde=True)  # + density curve
sns.histplot(data=tips, x="total_bill", hue="time", multiple="stack")   # layer, stack, dodge, fill
sns.kdeplot(data=tips, x="total_bill", hue="time", fill=True)          # smooth density
sns.ecdfplot(data=tips, x="total_bill")                     # cumulative
sns.rugplot(data=tips, x="total_bill")                      # ticks per value
sns.displot(data=tips, x="total_bill", col="time", kde=True)            # figure-level, facets

6. Categorical Plots

Comparing a numeric value across categories. Counts (countplot), estimates (barplot), distributions (boxplot, violinplot), points (stripplot).

Use it to answer questions like "Which day has the highest bills?", "how do salaries differ by department?".

One category axis and one numeric axis.

# Counts
sns.countplot(data=tips, x="day")                           # rows per category
sns.countplot(data=tips, x="day", hue="sex")

# Estimate (mean + confidence interval)
sns.barplot(data=tips, x="day", y="total_bill")             # mean with error bar
sns.barplot(data=tips, x="day", y="total_bill", estimator="sum", errorbar=None)
sns.pointplot(data=tips, x="day", y="total_bill", hue="sex")

# Distribution per category
sns.boxplot(data=tips, x="day", y="total_bill")             # median, quartiles, outliers
sns.violinplot(data=tips, x="day", y="total_bill", hue="sex", split=True)
sns.boxenplot(data=tips, x="day", y="total_bill")           # for large data

# Every point
sns.stripplot(data=tips, x="day", y="total_bill", jitter=True)
sns.swarmplot(data=tips, x="day", y="total_bill")           # no overlap

# Combine: box + points
sns.boxplot(data=tips, x="day", y="total_bill", color="lightgray")
sns.stripplot(data=tips, x="day", y="total_bill", size=3)

# Figure-level version of all of the above
sns.catplot(data=tips, x="day", y="total_bill", kind="box", col="time")

Horizontal: swap x and y (category on y).

7. Relationship Plots

Relationship between two numeric variables. scatterplot for points, lineplot for trends (with confidence band).

Use it for correlation checks, time series by group.

Two numeric variables.

sns.scatterplot(data=tips, x="total_bill", y="tip")
sns.scatterplot(data=tips, x="total_bill", y="tip", hue="day", size="size", style="time")
sns.lineplot(data=df, x="month", y="sales")                 # mean + confidence band
sns.lineplot(data=df, x="month", y="sales", hue="region", errorbar=None, marker="o")
sns.relplot(data=tips, x="total_bill", y="tip", col="time", hue="smoker")   # figure-level
sns.relplot(data=df, x="month", y="sales", kind="line")

8. Regression Plots

Scatter plot with a fitted trend line. regplot / lmplot fit a linear (or polynomial) model and draw it.

Use it for quick visual check of a linear relationship before building a model.

sns.regplot(data=tips, x="total_bill", y="tip")             # scatter + fit line
sns.regplot(data=tips, x="total_bill", y="tip", ci=None, scatter_kws={"alpha": 0.5})
sns.regplot(data=tips, x="total_bill", y="tip", order=2)    # polynomial fit
sns.lmplot(data=tips, x="total_bill", y="tip", hue="smoker", col="time")   # figure-level
sns.residplot(data=tips, x="total_bill", y="tip")           # residuals

9. Heatmap and Correlation

Colour-coded matrix of values. sns.heatmap(matrix, annot=True); often on df.corr() or a pivot table.

Use it for correlation overview of many columns, or a two-category summary table.

corr = tips.corr(numeric_only=True)
sns.heatmap(corr, annot=True, fmt=".2f", cmap="coolwarm", vmin=-1, vmax=1, center=0)

# Pivot then heatmap
table = tips.pivot_table(values="tip", index="day", columns="time", aggfunc="mean")
sns.heatmap(table, annot=True, cmap="Blues", linewidths=0.5)

sns.clustermap(corr, cmap="coolwarm")                       # clustered heatmap

10. Pair Plot and Joint Plot

Many relationships in one view. pairplot draws every pair of numeric columns; jointplot one pair with margins.

Use it for a first exploration of a new dataset with several numeric columns.

sns.pairplot(iris, hue="species")                           # every numeric pair
sns.pairplot(iris, hue="species", corner=True, diag_kind="kde")
sns.pairplot(tips, vars=["total_bill", "tip", "size"])

sns.jointplot(data=tips, x="total_bill", y="tip")           # scatter + both histograms
sns.jointplot(data=tips, x="total_bill", y="tip", kind="reg")   # scatter, kde, hist, hex, reg, resid

11. Figure-level vs Axes-level

The two kinds of seaborn functions and how they differ. Axes-level draw into one ax; figure-level create their own figure and support facets.

Use it for deciding between ax= (combine with subplots) and col= / row= (facets).

Figure-level (own figure, supports col / row) Axes-level (draws into one ax)
displot histplot, kdeplot, ecdfplot, rugplot
catplot countplot, barplot, boxplot, violinplot, stripplot, swarmplot, pointplot
relplot scatterplot, lineplot
lmplot regplot
pairplot, jointplot - 
# Axes-level: put into matplotlib subplots
fig, axes = plt.subplots(1, 2, figsize=(12, 4))
sns.histplot(data=tips, x="tip", ax=axes[0])
sns.boxplot(data=tips, x="day", y="tip", ax=axes[1])

# Figure-level: set size with height / aspect (not figsize)
g = sns.catplot(data=tips, x="day", y="tip", kind="box", height=4, aspect=1.5)
g.set_axis_labels("Day", "Tip")
g.set_titles("{col_name}")
g.figure.suptitle("Tips by Day", y=1.03)

12. Facets (Grid of Plots)

A grid of the same plot split by categories. col= / row= in figure-level functions, or FacetGrid for full control.

Use it for comparing a pattern across groups (lunch vs dinner, smokers vs non-smokers).

sns.relplot(data=tips, x="total_bill", y="tip", col="time", row="smoker")
sns.catplot(data=tips, x="day", y="tip", kind="bar", col="sex", col_wrap=2)

g = sns.FacetGrid(tips, col="time", row="sex")
g.map_dataframe(sns.histplot, x="tip")
g.add_legend()

13. Colors and Palettes

Choosing colours. Named palettes via palette= or set_palette; match the palette type to the data.

Use it for categories (qualitative), low-to-high (sequential), around zero (diverging).

sns.color_palette()                                 # current palette
sns.set_palette("Set2")                             # default for all plots
sns.boxplot(data=tips, x="day", y="tip", palette="pastel", hue="day", legend=False)
sns.scatterplot(data=tips, x="total_bill", y="tip", hue="size", palette="viridis")
sns.barplot(data=tips, x="day", y="tip", color="steelblue")    # one color
Type Use for Examples
Qualitative categories deep, muted, pastel, Set2, tab10, colorblind
Sequential low to high Blues, viridis, rocket, mako
Diverging negative / zero / positive coolwarm, vlag, RdBu, icefire

14. Customize with Matplotlib

Fine-tuning seaborn charts with matplotlib. Axes-level functions return ax; use ax.set_title, ax.bar_label and so on.

Use it for titles, rotated labels, value labels, moving the legend.

Axes-level functions return a matplotlib ax:

ax = sns.barplot(data=tips, x="day", y="tip")
ax.set_title("Average Tip by Day")
ax.set_xlabel("") ; ax.set_ylabel("Tip (USD)")
ax.tick_params(axis="x", rotation=45)
ax.bar_label(ax.containers[0], fmt="%.2f")         # value on bars
sns.move_legend(ax, "upper left", bbox_to_anchor=(1, 1))   # legend outside
plt.tight_layout()
plt.show()

See 21 - Matplotlib for more options.

15. Save a Figure

Writing seaborn charts to files. ax.figure.savefig (axes-level) or g.savefig (figure-level).

Use it for charts for reports, slides and READMEs.

# Axes-level
ax = sns.histplot(data=tips, x="tip")
ax.figure.savefig("tips.png", dpi=300, bbox_inches="tight")

# Figure-level
g = sns.pairplot(iris, hue="species")
g.savefig("pairplot.png", dpi=300)

16. Which Plot to Use

A lookup table from question to plot type. Find your question on the left, use the plot on the right.

Use this when you know what you want to show but not which chart fits.

Question Plot
How is one numeric column distributed? histplot, kdeplot, boxplot
How many rows per category? countplot
Compare an average across categories barplot, pointplot
Compare distributions across categories boxplot, violinplot, stripplot
Relationship between two numeric columns scatterplot, regplot, jointplot
Trend over time lineplot
Correlation between many columns heatmap of df.corr(), pairplot
Same plot split by a category col= / row= in relplot, catplot, displot

17. Troubleshooting

Problem Fix
Nothing appears (script) Add plt.show()
figsize has no effect Figure-level function: use height= and aspect=
ax= not accepted Figure-level functions (catplot, relplot...) create their own figure; use the axes-level version
FutureWarning: Passing palette without hue Add hue= (same column as x) and legend=False
ValueError: could not convert string to float in heatmap Use df.corr(numeric_only=True) or select numeric columns
Plots stack on top of each other Call plt.figure() or plt.subplots() before each new plot
load_dataset fails Needs internet; use your own DataFrame
Categories in wrong order order=["Thur", "Fri", "Sat", "Sun"]

18. Try It

Short exercises to practise this guide. Try each task yourself first, then open the solution.

Use it right after reading the guide, or later as a quick self-test.

Exercise 1: Distribution by group

Histogram of total_bill from the tips dataset, coloured by time, with a density curve.

Solution 
tips = sns.load_dataset("tips")
sns.histplot(data=tips, x="total_bill", hue="time", kde=True)

Exercise 2: Ordered boxplot

Box plot of tip per day in the order Thur, Fri, Sat, Sun.

Solution 
sns.boxplot(data=tips, x="day", y="tip", order=["Thur", "Fri", "Sat", "Sun"])

Exercise 3: Correlation heatmap

Annotated heatmap of the correlations of the numeric tips columns.

Solution 
sns.heatmap(tips.corr(numeric_only=True), annot=True, fmt=".2f", cmap="coolwarm", vmin=-1, vmax=1)

Previous: 21 - Matplotlib | Index: All guides | Next: 23 - Scikit-learn