Skip to main content

Introduction to Jupyter Notebooks

Jupyter Notebook combines executable cells, their outputs, and Markdown narrative. A notebook's visible cell order is not proof of execution order, so reproducibility requires deliberate kernel and execution practices.

Install and launch​

Create and activate a virtual environment, then install the classic Notebook interface:

python -m pip install notebook
jupyter notebook

JupyterLab is a separate interface installed with python -m pip install jupyterlab and launched with jupyter lab.

Confirm the kernel​

A notebook executes through its selected kernel. In a code cell, inspect the interpreter:

import sys

print(sys.executable)
print(sys.version.split()[0])

The path should identify the intended environment. Installing a package into a terminal's Python does not guarantee that a different notebook kernel can import it.

Cells and execution state​

  • Shift+Enter runs a cell and moves to the next one.
  • Ctrl+Enter runs a cell and stays on it.
  • Markdown cells hold explanations, headings, equations, and links.
  • Code cells share kernel state until the kernel restarts.

This example is deterministic when run after a fresh kernel start:

values = [2, 4, 8]
mean = sum(values) / len(values)
print(mean)  # 4.666666666666667

Before sharing a notebook, restart the kernel and run all cells from top to bottom. Remove secrets and unnecessary large outputs, keep data-loading assumptions visible, and record dependency versions outside the notebook in the project environment declaration.

Notebook trust​

Notebook files may contain stored HTML, JavaScript-capable output, and code. Open untrusted notebooks cautiously, inspect code before running it, and do not treat a rendered output as evidence that you executed the displayed source yourself.