Seaborn Intermediate¶

What you'll learn¶

• What Seaborn adds over Matplotlib
• Statistical plots in one line
• The main plot categories
• Working with DataFrames
• When to use Seaborn vs Matplotlib

Seaborn is a statistical visualization library built on top of Matplotlib. It makes attractive, informative statistical graphics with far less code — and understands Pandas DataFrames directly.

What Seaborn adds¶

A core question explored in Seaborn: What Seaborn adds.

Matplotlib is powerful but low-level — a statistical plot can take many lines. Seaborn wraps common statistical visualizations into single calls, adds attractive defaults, and speaks DataFrames:

import seaborn as sns          # pip install seaborn
import matplotlib.pyplot as plt

tips = sns.load_dataset("tips")            # a built-in example DataFrame

# A scatter plot with a regression line — one call
sns.regplot(data=tips, x="total_bill", y="tip")
plt.show()

The same plot in raw Matplotlib would need manual regression fitting and styling. Seaborn does it in a line, and it looks good by default.

The main plot categories¶

Relational, distribution, categorical, and matrix plots — the Seaborn families.

Seaborn organizes plots by what they show:

sns.histplot(data=tips, x="total_bill", kde=True)   # distribution + density curve
sns.boxplot(data=tips, x="day", y="total_bill")      # distribution per category
sns.heatmap(tips.corr(numeric_only=True), annot=True)  # correlation matrix

The killer features¶

Built-in statistical estimation, faceting, and themes that Matplotlib lacks.

Two Seaborn capabilities are especially powerful:

pairplot — plot every pair of numeric columns at once, instantly revealing relationships across a whole dataset:

sns.pairplot(tips, hue="time")   # scatter matrix, colored by a category

Semantic mapping with hue/size/style — encode extra dimensions by color, size, or marker, so one plot shows three or four variables:

sns.scatterplot(data=tips, x="total_bill", y="tip",
                hue="day", size="size", style="smoker")

These turn exploratory data analysis into a few expressive lines — a big reason Seaborn is a staple of data science notebooks.

Seaborn vs Matplotlib¶

Seaborn for fast statistical plots; drop to Matplotlib for fine control.

They're complementary, not competitors (Seaborn is Matplotlib underneath):

• Seaborn — for statistical plots, quick exploration, attractive defaults, DataFrame-native work. Reach for it first in data analysis.
• Matplotlib — for fine-grained control, custom/non-statistical plots, and final polish. You often start with Seaborn and tweak with Matplotlib (plt. calls work on Seaborn plots).

Practice exercises¶

1. Describe which Seaborn plot you'd use to see the distribution of a single numeric column.
2. Explain what hue="category" adds to a scatter plot and why it's useful.
3. When would you use a boxplot vs a violinplot for showing distributions per category?
4. Describe what a heatmap of a correlation matrix reveals at a glance.
5. Explain the division of labor between Seaborn and Matplotlib in a real analysis workflow.