Course contentsShow
← Data Science
Lesson 2049 of 2,14547. Reproducible ResearchFree lesson

Requirements Files in Python

Creating and using requirements.txt to explicitly list all packages and versions needed to run your analysis.

Requirements Files in Python

You'll learn how to create and use requirements.txt files to explicitly document every package and version your analysis needs, ensuring anyone can recreate your exact environment.

What Is a Requirements File?

A requirements file (typically named requirements.txt) is a plain text document that lists all the Python packages your project depends on, along with their specific versions. Think of it as a recipe card that tells other chefs exactly which ingredients and measurements they need to cook your dish perfectly.

When you've installed dozens of packages over weeks of experimentation, a requirements file captures that precise configuration so you—or anyone else—can recreate the same working environment months or years later.

Creating a Requirements File

If you're working in an activated virtual environment (covered in previous lessons), you can generate a requirements file automatically:

pip freeze > requirements.txt

This captures every installed package with exact version numbers:

pandas==1.5.3
numpy==1.24.2
scikit-learn==1.2.1
matplotlib==3.7.0

Using a Requirements File

To install all dependencies listed in a requirements file:

pip install -r requirements.txt

This single command reads the file and installs every package at the specified version, recreating your analysis environment on a colleague's machine, a production server, or your own computer after a fresh setup.

Why Exact Versions Matter

Without version pinning (==1.5.3), pip installs the latest available version, which may introduce breaking changes or different numerical results. In reproducible research, exact versions guarantee identical behavior across environments and time.

Key Takeaway: A requirements.txt file is your project's dependency snapshot—create it with pip freeze, share it with your code, and use pip install -r to guarantee reproducible environments every time.