Tutorial 5: Data Visualization with Seaborn -

Introduction

Seaborn is a Python library built on top of Matplotlib. It makes statistical visualization easier and produces attractive, informative charts with far less code.

Matplotlib → Basic plotting control
Seaborn → Statistical visualization → Patterns • Relationships • Distributions

You’ll learn to create: count plots, bar plots, histograms and density plots, box plots, scatter plots, heatmaps, and pair plots.

1. Install and Import Seaborn

If not already installed:

pip install seaborn

In a Jupyter Notebook:

%pip install seaborn

Import the libraries:

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

(Seaborn is conventionally imported as sns — a nod to the character Samuel Norman Seaborn from The West Wing, which is where the library’s creator got the name.)

2. Create a Sample Dataset

data = {
    "Name": ["Amit", "Priya", "Rahul", "Sneha", "Vikram", "Anjali"],
    "Department": ["IT", "HR", "Finance", "IT", "Sales", "Finance"],
    "Salary": [75000, 60000, 85000, 90000, 65000, 95000],
    "Experience": [5, 3, 7, 8, 4, 10]
}

df = pd.DataFrame(data)
df

3. Count Plot

A count plot shows how many records fall into each category.

sns.countplot(data=df, x="Department")
plt.title("Employees by Department")
plt.xlabel("Department")
plt.ylabel("Number of Employees")
plt.show()

Use it for: department → number of employees, city → number of customers, category → number of products.

4. Bar Plot

A bar plot compares a numerical value across categories — for example, average salary by department.

sns.barplot(data=df, x="Department", y="Salary")
plt.title("Average Salary by Department")
plt.show()

Seaborn automatically calculates an aggregate (by default, the mean) for each category. You can also specify it explicitly:

sns.barplot(data=df, x="Department", y="Salary", estimator="mean")

5. Histogram

sns.histplot(data=df, x="Salary")
plt.title("Salary Distribution")
plt.show()

You can add a smooth density curve on top:

sns.histplot(data=df, x="Salary", kde=True)
plt.title("Salary Distribution")
plt.show()

What is kde=True? KDE stands for Kernel Density Estimate — it draws a smoothed curve approximating the shape of the underlying distribution, which is often easier to interpret at a glance than raw histogram bars.

6. Box Plot

A box plot summarizes a numerical column using five key statistics: minimum, maximum, median, and the two quartiles (25th and 75th percentiles) — and it also visually flags outliers.

sns.boxplot(data=df, y="Salary")
plt.title("Salary Distribution")
plt.show()

Compare salary by department:

sns.boxplot(data=df, x="Department", y="Salary")
plt.title("Salary by Department")
plt.show()

7. Scatter Plot

sns.scatterplot(data=df, x="Experience", y="Salary")
plt.title("Experience vs Salary")
plt.show()

This helps investigate: does salary tend to increase as experience increases? You can add a trend line using regplot():

sns.regplot(data=df, x="Experience", y="Salary")
plt.title("Experience vs Salary")
plt.show()

8. `hue`: Adding a Third Dimension

The hue parameter colors data points by an additional categorical variable:

sns.scatterplot(data=df, x="Experience", y="Salary", hue="Department")
plt.title("Experience vs Salary by Department")
plt.show()

Conceptually: X-axis → Experience, Y-axis → Salary, Hue (color) → Department. This lets you analyze multiple dimensions in a single chart.

9. Heatmap

A heatmap represents values using color intensity — most commonly used to visualize correlations.

First calculate correlations:

correlation = df[["Salary", "Experience"]].corr()
correlation

Then visualize:

sns.heatmap(correlation, annot=True)
plt.title("Correlation Heatmap")
plt.show()

annot=True displays the actual numeric correlation value inside each cell. Correlation values range from -1 (strong negative relationship) through 0 (no linear relationship) to +1 (strong positive relationship).

10. Pair Plot

A pair plot automatically creates a grid of scatter plots (and histograms) for every pair of numerical columns:

sns.pairplot(df[["Salary", "Experience"]])
plt.show()

It’s an efficient way to quickly scan for relationships across many variables at once, without writing a separate scatter plot for each pair.

11. Seaborn Styling

Seaborn ships with built-in visual themes:

sns.set_theme(style="whitegrid")

Other common styles:

sns.set_theme(style="darkgrid")
sns.set_theme(style="white")
sns.set_theme(style="ticks")

Then create your chart as usual:

sns.barplot(data=df, x="Department", y="Salary")
plt.title("Average Salary by Department")
plt.show()

12. Matplotlib vs. Seaborn

Feature	Matplotlib	Seaborn
Main purpose	General visualization	Statistical visualization
Control	Very high	High
Code required	Often more	Often less
Default appearance	Basic	More polished
Statistical charts	Manual	Built-in
Works with Pandas	Yes	Yes

Simple rule of thumb: Matplotlib gives you control. Seaborn gives you convenience and built-in statistical visualizations. In practice, they’re almost always used together:

import seaborn as sns
import matplotlib.pyplot as plt

sns.barplot(data=df, x="Department", y="Salary")
plt.title("Average Salary by Department")
plt.show()

Example: Visualizing Department Data

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

data = {
    "Department": ["IT", "IT", "IT", "HR", "HR", "Finance", "Finance", "Finance", "Sales"],
    "Salary": [75000, 90000, 85000, 60000, 65000, 85000, 95000, 88000, 65000]
}

df = pd.DataFrame(data)

sns.barplot(data=df, x="Department", y="Salary")
plt.title("Average Salary by Department")
plt.xlabel("Department")
plt.ylabel("Average Salary")
plt.show()

Possible insight: Finance has the highest average salary among the departments in this dataset.

Choosing the Right Seaborn Chart

Need to count categories? → countplot
Compare numerical values? → barplot
Understand distribution? → histplot
Find outliers? → boxplot
Study relationships? → scatterplot
Study correlations? → heatmap
Explore many variables at once? → pairplot