Seaborn as a Higher-Level Plotting Library

Seaborn is a third-party plotting library built on top of matplotlib. The conventional import is:

import seaborn as sns

The examples and exercises in this chapter require Seaborn 0.12 or later. If the version check from the overview reports an older version, upgrade Seaborn in the same virtual environment:

python -m pip install --upgrade "seaborn>=0.12"

Matplotlib gives you the basic plotting building blocks. Seaborn gives you convenient functions for common statistical charts. It works with table-shaped data and can calculate statistical summaries and uncertainty intervals for some plots. You do not need to learn any new statistics for this chapter.

Seaborn works especially well with pandas DataFrames. Instead of passing separate lists, you often pass a DataFrame and name the columns:

sns.barplot(data=summary, x="route", y="kilometres", errorbar=None, ax=ax)

The column names are strings. That means they must match your real DataFrame columns.

A Bicycle-Ride DataFrame

This chapter uses a small log of bicycle rides, giving you a fresh context for the same prepare-then-plot pattern:

import pandas as pd

records = [
    {"route": "Riverside", "kilometres": 8, "date": "2026-05-10"},
    {"route": "Park loop", "kilometres": 5, "date": "2026-05-10"},
    {"route": "Riverside", "kilometres": 12, "date": "2026-05-11"},
    {"route": "Hill route", "kilometres": 7, "date": "2026-05-12"},
    {"route": "Park loop", "kilometres": 9, "date": "2026-05-13"},
    {"route": "Riverside", "kilometres": 6, "date": "2026-05-13"},
]

df = pd.DataFrame(records)

A Seaborn Bar Chart

Prepare the route totals:

summary = (
    df.groupby("route", as_index=False)["kilometres"]
    .sum()
    .sort_values("kilometres", ascending=False)
)

Then plot them:

import matplotlib.pyplot as plt
import seaborn as sns

fig, ax = plt.subplots()
sns.barplot(data=summary, x="route", y="kilometres", errorbar=None, ax=ax)
ax.set_title("Total cycling distance by route")
ax.set_xlabel("Route")
ax.set_ylabel("Kilometres")
plt.show()

The chart still uses matplotlib’s fig and ax. Seaborn draws onto the axes you provide. Because summary already contains exactly one total per route, errorbar=None turns off uncertainty intervals that would not add useful information here.

df = pd.DataFrame([
    {"team": "Red", "points": 12, "date": "2026-05-10"},
    {"team": "Blue", "points": 9, "date": "2026-05-10"},
    {"team": "Red", "points": 6, "date": "2026-05-11"},
    {"team": "Green", "points": 14, "date": "2026-05-11"},
    {"team": "Blue", "points": 16, "date": "2026-05-12"},
])

A Seaborn Line Chart

Prepare daily cycling distances:

daily = df.copy()
daily["date"] = pd.to_datetime(daily["date"])
daily = daily.groupby("date", as_index=False)["kilometres"].sum()
daily = daily.sort_values("date")

Then plot them:

fig, ax = plt.subplots()
sns.lineplot(data=daily, x="date", y="kilometres", marker="o", ax=ax)
ax.set_title("Cycling distance over time")
ax.set_xlabel("Date")
ax.set_ylabel("Kilometres")
plt.show()

The same rule applies: prepare the data first. Seaborn makes the plotting call shorter, but it does not decide what question you are asking.

A Seaborn Histogram

For the distribution of individual ride distances:

fig, ax = plt.subplots()
sns.histplot(data=df, x="kilometres", bins=5, ax=ax)
ax.set_title("Distribution of ride distances")
ax.set_xlabel("Kilometres")
ax.set_ylabel("Number of rides")
plt.show()

The DataFrame already has one row per ride, so no grouping is needed. The chart summarizes the kilometres column into ranges.

What Seaborn Does Not Solve

Seaborn can make a chart look cleaner by default. It can also save you some matplotlib details. But it cannot know your project contract.

If your DataFrame columns are:

route, kilometres, date

then this code is wrong for your data:

sns.barplot(data=df, x="trail", y="distance")

It may be a normal seaborn pattern for some other table. It is not correct for this bicycle-ride log.

Use the same library-checking habit from Using Libraries:

print(df.columns)
print(df.head())

Then write the seaborn call with the real columns.

Check Your Understanding

1. How does seaborn relate to matplotlib?
2. What does sns.barplot(data=summary, x="route", y="kilometres", errorbar=None, ax=ax) expect to find in summary?
3. Why does seaborn not remove the need to prepare a summary first?
4. How can you check that a seaborn example from documentation matches your DataFrame?