Jupyter Notebooks¶

Jupyter notebooks let you mix code, text, and visualizations in a single document. They're the standard tool for data exploration, prototyping, and sharing analysis, widely used in data science, machine learning, and scientific computing.

Learn Your Way¶

Read	Build	Watch	Test	Review	Visualize
You are here	Projects	—	Quiz	Flashcards	—

Why Jupyter Matters¶

When you're exploring data or testing ideas, you don't want to write a full script, run it, tweak it, and run it again. Notebooks let you run code in small chunks (cells) and see the output immediately, including charts, tables, and images. This interactive workflow is why data scientists live in Jupyter.

Installing Jupyter¶

# JupyterLab (recommended — modern interface)
pip install jupyterlab
jupyter lab

# Classic Notebook (simpler, still widely used)
pip install notebook
jupyter notebook

Both open in your browser at http://localhost:8888.

Notebook Basics¶

A notebook (.ipynb file) is a sequence of cells. Each cell is either:

Code Cells¶

# This runs Python and shows output below the cell
import pandas as pd

df = pd.read_csv("data.csv")
df.head()  # Output appears directly below

Press Shift+Enter to run a cell and move to the next one.

Markdown Cells¶

Write formatted text using Markdown:

# Section Title
This is a **bold** explanation with a [link](https://docs.python.org).

- Bullet points work
- So do numbered lists

Use markdown cells to document your analysis as you go.

Magic Commands¶

Jupyter has special commands prefixed with % (line) or %% (cell):

# Time how long a line takes
%timeit sorted(range(1000))

# Time an entire cell
%%timeit
data = list(range(10000))
sorted(data)

# Run a shell command
!pip install requests

# Show matplotlib plots inline
%matplotlib inline

# List all variables in the namespace
%whos

Common Workflow¶

# Cell 1: Imports
import pandas as pd
import matplotlib.pyplot as plt

# Cell 2: Load data
df = pd.read_csv("sales.csv")

# Cell 3: Explore
df.describe()

# Cell 4: Visualize
df.plot(x="month", y="revenue", kind="bar")
plt.title("Monthly Revenue")
plt.show()

# Cell 5: Analysis
top_month = df.loc[df["revenue"].idxmax()]
print(f"Best month: {top_month['month']} (${top_month['revenue']:,.0f})")

Each cell builds on the previous ones. You can go back and re-run any cell.

Exporting Notebooks¶

# To Python script
jupyter nbconvert --to script analysis.ipynb

# To HTML (for sharing)
jupyter nbconvert --to html analysis.ipynb

# To PDF (requires pandoc + LaTeX)
jupyter nbconvert --to pdf analysis.ipynb

Notebooks vs Scripts¶

Use Notebooks When...	Use Scripts When...
Exploring data interactively	Building reusable tools
Prototyping and experimenting	Running in production
Creating shareable analysis	Writing library code
Teaching or presenting	Working in CI/CD pipelines
Quick visualizations	Code that others import

Rule of thumb: Start in a notebook, then extract reusable code into .py files.

Common Mistakes¶

Running cells out of order. Notebooks execute cells in whatever order you click them. If you define x = 5 in cell 3 but run cell 5 first, x won't exist. Always use "Restart & Run All" before sharing.

Hidden state. You might delete a cell that defined a variable, but the variable still exists in memory. "Restart & Run All" catches this.

Notebooks in version control. .ipynb files are JSON with embedded output (images, data). They create messy git diffs. Solutions: - Clear output before committing - Use nbstripout to auto-strip output on commit - Use jupytext to keep notebooks as .py files

JupyterLite: Browser-Based¶

JupyterLite runs entirely in your browser, no installation needed. It uses Pyodide (Python compiled to WebAssembly). Great for: - Quick experiments without setup - Sharing interactive examples - Teaching in environments where installing software is difficult

Practice This¶

Open JupyterLite and run import this (The Zen of Python)
Create a notebook that loads a CSV and makes a chart
Try the Module 07 data analysis projects in a notebook
Export a notebook to HTML and open it in a browser