Notebooks

Introduction

As for now, we mostly worked with .py files - regular python script files with a plain text containing the code. This is a golden standart of Python development that can be met everywhere: from software development (web, games, and so on) to machine learning and AI stuff.

That being said, this is not the only file format to work with Python. Today we will briefly introduce you to Notebooks, specifically - .ipynb Jupyter Notebooks.

(src: jupyter project site)

What is and Why Notebooks?

tl;dr:

Advantages of Notebooks:

✔︎ Great for data analysis, visualization, and documentation.
✔︎ Supports markdown, making it easy to explain code.
✔︎ Can execute code step-by-step - no need to rerun the whole script!

Disadvantages of Notebooks:

✘ Hidden state creates serious barriers for novice programmers.
✘ Harder to maintain, manage dependencies and version control.
✘ Overall considered as non convenient for project development.

.py

Imagine: You enjoy Python and working with data, so you write scripts to automate a routine analysis. It works well - check results as you go, or show your results to someone unfamiliar with your analysis? In this way, scripts are not the most convenient way to organise the code.

.ipynb

On the other hand, Notebooks are documents that contain both computer code (not only Python!) and rich text elements (paragraph, equations, figures, links, etc). Combining code with visual output adds valuable context for the reader. Since code is usually read more often than it’s written, having explanations and charts will make it easier to understand your thought process - even weeks later. This is especially important for working with real-world data, where data specificity shapes your code, and presenting your outcomes.

At the highest level, a Jupyter notebook is a dictionary with a few keys:

• metadata (dict)
• nbformat (int)
• nbformat_minor (int)
• cells (list)

However, usage of Notebooks is pretty limited. To put it simply: they’re a maintainability nightmare. You copypaste code from cell to cell, making it difficult to debug and modularize. The files size is bigger (especially with images), jsons are harder for version control. Another thing that can create unnecessary barriers is hidden state: because users can run code in any order, this can sometimes lead to inconsistencies: where the recorded output no longer matches the actual results when running cells in sequence, or calling variable before defining it.

To sum up: it’s more of a sharing/ presentation tool than anything else. But it’s great for that.

How to Run Jupyter Notebook

There are multiple ways how you can try out working with Notbooks.

1. Web-Based Jupyter Notebook IDEs (Easiest)
   Work on your project without any setup or specific hardware, but you’ll be tied to proprietary software. Popular options:
   • Google Colab
   • JupyterLab (Online)

2. Using Anaconda (GUI-Based)
   A beginner-friendly way to run Jupyter locally. For this you need to:
   1. Install Anaconda distribution platform for Python
   2. Open Anaconda Navigator.
   3. Click Launch on Jupyter Notebook.

3. Using VS Code extentions
   Great for integrating notebooks into a comfortable coding environment.
   • Install Python and Jupyter extensions.
   • Open a .ipynb file and run cells directly.

4. Using Terminal or Command Prompt A more flexible approach, requiring manual installation.
   1. Install Jupyter: pip install jupyter
   2. Start Jupyter: jupyter notebook