You’ve hit the limits of your spreadsheet. Maybe the file has 2 million rows and keeps crashing. Maybe you rebuild the same monthly report by hand, clicking through the same 40 steps. Or maybe a colleague sent you a file ending in .ipynb and you’re not sure what to do with it.

That file is a Jupyter notebook, and it’s the most common way analysts take their first steps with Python. This guide explains what a notebook is, how to get one running in a few minutes and what your first real session should look like if you’re coming from Excel or Google Sheets.

What a Jupyter notebook is

A Jupyter notebook is a document that combines 4 things in one place:

• Code you can run, usually Python
• Outputs from that code, such as tables, numbers and error messages
• Charts and other visuals
• Written notes in Markdown, such as headings, explanations and bullet points

Project Jupyter’s documentation describes a notebook as a document that mixes explanatory text, code, images and outputs. You run the code in small pieces called cells, and each result appears directly below its cell.

The name comes from the 3 languages the project originally supported: Julia, Python and R. Python is by far the most common choice for business analysis today. The project grew out of IPython, an enhanced interactive Python shell that still powers the default Python kernel.

Notebooks are saved as .ipynb files. Under the hood, these are structured JSON documents that store your cells, their outputs and some metadata. That’s why opening one in a plain text editor looks messy, but opening it in a notebook tool looks clean.

How notebooks work: cells and kernels

2 ideas make notebooks click.

Cells. A notebook is a list of cells. A code cell holds code you can run. A Markdown cell holds formatted text. You run a cell with Shift + Enter, see the result and move to the next one.

The kernel. Behind every open notebook is a kernel, a separate process that runs your code and remembers your variables. Load a file in cell 1, and the data stays in memory for cells 2, 3 and 4. Restart the kernel and that memory is wiped.

This is the biggest mental shift from spreadsheets. In Excel, change a cell and every formula recalculates. In a notebook, nothing reruns unless you run it. If you edit cell 2 and skip running it, cell 5 still uses the old result.

Key takeaway: Treat a notebook like a recipe, top to bottom. Before you share one, use “Restart kernel and run all cells” to prove every step still works in order.

Spreadsheet vs notebook: when to use which

Notebooks don’t replace spreadsheets. They solve different problems.

Task	Spreadsheet	Jupyter notebook
Quick edits and small lists	Best choice	Overkill
Sharing with non-technical colleagues	Easy	Harder, needs export
Files over about 1 million rows	Excel caps a sheet at 1,048,576 rows	Handles well with pandas
Repeating the same cleanup every month	Manual or macros	Rerun the notebook
Joining several files or data sources	Lookups get fragile	Built-in merge tools
Showing your work step by step	Hard to audit	Each step is visible
Charts for a quick look	Fast	Fast, more control
Statistics and forecasting	Limited	Full Python libraries

If you’re still deciding between languages, our comparison of Python vs R for data analysis covers the trade-offs.

3 ways to open your first notebook

You don’t need to be a developer to start. Pick the option that fits your situation.

Option	Setup	Cost	Best for
Google Colab	None, runs in your browser	Free tier, paid upgrades	Trying notebooks today without installing anything
JupyterLab or Jupyter Notebook on your computer	Install Python, then one command	Free, open source	Working with local or sensitive files
VS Code with the Jupyter extension	Install VS Code, Python and the extension	Free	People who also write scripts or want one editor for everything

Google Colab

Colab is a hosted Jupyter notebook service from Google. Its FAQ says it’s free to use, stores notebooks in Google Drive and is built on Jupyter. The catch: free resources are not guaranteed, idle sessions time out and free notebooks run for at most 12 hours. Check your company’s policy before uploading business data to any cloud tool.

JupyterLab or Jupyter Notebook locally

Once Python is installed, the official Jupyter install page gives 2 one-line options:

• pip install jupyterlab, then jupyter lab
• pip install notebook, then jupyter notebook

Either opens a tab in your browser running on your own machine. Jupyter Notebook is the simpler interface, good for focused work and learning. JupyterLab adds tabs, side-by-side panels, a file browser and a terminal. Current versions of Jupyter Notebook (version 7 and later) are built from JupyterLab components, so the 2 feel familiar.

VS Code

Visual Studio Code supports notebooks through its Jupyter extension. You select a Python environment that has Jupyter installed, pick a kernel and run cells inside the editor.

Your first session: from spreadsheet to Python

Here’s a realistic first hour for an analyst. The goal is to repeat something you already do in Excel, so you can compare results.

Step 1: Set up the notebook

Create a new notebook and add a Markdown cell at the top with a title, the date and one sentence on what the notebook does. Future you will thank you.

Step 2: Import your tools

In the first code cell, import pandas (for tables) and matplotlib (for charts):

python import pandas as pd import matplotlib.pyplot as plt

If you get an error that a module isn’t found, install it with pip install pandas matplotlib in a terminal, or !pip install pandas matplotlib in a notebook cell.

Step 3: Load a file

```python sales = pd.read_csv(“sales.csv”)

or for Excel:

sales = pd.read_excel(“sales.xlsx”) ```

For .xlsx files, pandas uses the openpyxl library by default, so install it if you see an error.

Step 4: Look before you calculate

python sales.head() sales.info() sales.describe()

These 3 lines show the first rows, the column types and missing values, and basic statistics. It’s the notebook version of scrolling through a sheet and eyeballing it. Our guide on how to clean data covers what to fix next.

Step 5: Recreate a pivot table

python sales.groupby("region")["revenue"].sum().sort_values(ascending=False)

That one line does what a basic pivot table does. If you know pivot tables well, our Excel pivot tables tutorial is a helpful mental bridge.

Step 6: Make a chart

python sales.groupby("region")["revenue"].sum().plot(kind="bar") plt.show()

Step 7: Save and rerun

Save the notebook, then choose Restart kernel and run all. If everything runs top to bottom without errors, you now have a repeatable report. Next month, swap in the new file and run it again.

Good habits that save you later

• Name notebooks clearly, for example 2026-10-sales-by-region.ipynb.
• Keep raw data untouched. Load it, then create cleaned versions in code. Never overwrite the original file.
• Use Markdown cells to explain why you did something, not just what.
• Keep cells short. One step per cell makes errors easy to find.
• Don’t hard-code secrets. Never paste passwords or API keys into a notebook you might share.
• Use version control once notebooks matter. Our Git for beginners guide is a good start.
• Clear sensitive outputs before sharing, since outputs are saved inside the file.

Where to go next

Once you’re comfortable loading, summarizing and charting data, the next skills are:

1. SQL, to pull data straight from databases. Start with SQL for beginners.
2. Statistics basics, so you can tell real changes from noise. See statistics basics for data analysis.
3. Automation, by turning a working notebook into a scheduled report.

If you’re building toward an analyst role, our guide on how to become a data analyst maps the full skill path. For terms you’re unsure about, check the glossary.

Conclusion

A Jupyter notebook is a place to write code, see results and explain your thinking in one file. For analysts coming from spreadsheets, it’s the gentlest route into Python. Start in Colab or a local install, repeat a report you already know and rerun it from top to bottom. Once that works, you’ve built your first piece of repeatable analysis.

Sources

• Project Jupyter documentation: What is Jupyter? — notebooks, kernels, .ipynb JSON format, JupyterLab vs Notebook, name origin and IPython
• Project Jupyter: Installing Jupyter — install and launch commands
• Jupyter Notebook documentation: Migrating to Notebook 7 — Notebook 7 built on JupyterLab components
• Google Colab FAQ — hosted Jupyter service, free tier limits and storage in Google Drive
• Microsoft Support: Excel specifications and limits — 1,048,576-row worksheet limit
• pandas documentation: read_excel — openpyxl as the default engine for .xlsx files