What Is Jupyter Notebook? A Beginner's Guide for Data Science and ML

Jupyter Notebook is widely used in the fields of machine learning and data science. Many of you may have encountered the term “Jupyter compatible” when trying to use Google Colab. This article provides a guide to Jupyter Notebook, from its history to the latest usage methods, explained clearly for beginners.

What is Jupyter Notebook?

Jupyter Notebook is a web application that provides an interactive computing environment. Its major feature is the ability to integrate code, execution results, explanatory text, graphs, and mathematical formulas into a single document¹.

In traditional programming environments, code was written in text editors, executed in separate consoles, and results were checked in yet another location. Jupyter Notebook is an innovative tool that allows all of these to be completed on a single screen.

Key Features

1. Cell-based Execution

Write and execute code in small units (cells)
Build programs through trial and error
Fix and re-execute only the problematic parts when errors occur

2. Rich Output Formats

Direct display of graphs, tables, images, and videos
Interactive widgets available
Visual confirmation of results on the spot

3. Multi-language Support

Supports over 40 programming languages
Covers major languages like Python, R, Julia, and Scala
Switch languages using execution engines called kernels

History and Evolution

Beginning with IPython (2001~)

The history of Jupyter Notebook dates back to IPython, developed in 2001 by Fernando Pérez as a physics graduate student². The Python interpreter at that time lacked functionality for scientific computing, creating a need for a more powerful interactive environment.

Birth of Notebook Feature (2011)

In 2011, a team consisting of Fernando Pérez, Brian Granger, and Min Ragan-Kelley added a web-based Notebook feature to IPython. This became the prototype for the current Jupyter Notebook.

Evolution to Project Jupyter (2014)

In 2014, it was announced that the language-agnostic parts would be separated from the IPython project and made independent as Project Jupyter³. The name “Jupyter” is a combination of the main supported languages Julia, Python, and R, and also pays homage to Galileo’s discovery of Jupiter’s moons recorded in his notebooks.

Rapid Adoption and Recognition (2015~Present)

2015: About 200,000 Jupyter notebooks published on GitHub
2018: Increased to about 2.5 million
2021: Reached about 10 million
2021: Selected by Nature as one of “10 computing projects that transformed science”

Comparison with Google Colab

Google Colaboratory (Colab) is a cloud-based service based on Jupyter Notebook. Understanding the differences between them helps in selecting the appropriate tool.

Major Differences

Item	Jupyter Notebook	Google Colab
Execution Environment	Local (your PC)	Cloud (Google’s servers)
Setup	Installation and environment setup required	Immediately available with just a browser
Computing Resources	Depends on your PC specs	Free GPU/TPU available
Data Storage	Local storage	Google Drive
Collaboration	Basically individual use	Real-time collaborative editing possible
Internet	Works offline	Constant connection required
Customizability	High (extensions, settings)	Limited
Pricing	Completely free	Basic free (Pro version from $9.99/month)

Usage Guidelines

When Jupyter Notebook is Suitable:

Research and development handling confidential data
Processing large datasets
Need for custom environments or special libraries
Working in offline environments
Projects requiring complete control

When Google Colab is Suitable:

Quick prototyping
Deep learning experiments using GPU/TPU
Team collaboration
Educational and learning purposes
Want to avoid environment setup hassle

Applications in Machine Learning and AI

1. Data Exploration and Preprocessing

Jupyter Notebook is ideal for Exploratory Data Analysis (EDA) in data science workflows.

# Loading data and checking basic statistics
import pandas as pd
import matplotlib.pyplot as plt

df = pd.read_csv('data.csv')
df.describe()  # Display basic statistics immediately
df.plot()      # Display graphs inline

2. Model Development and Experimentation

In machine learning model development, you can interactively adjust parameters and compare methods.

# Comparing different models
from sklearn.model_selection import train_test_split
from sklearn.ensemble import RandomForestClassifier
from sklearn.svm import SVC

# Split data
X_train, X_test, y_train, y_test = train_test_split(X, y)

# Model 1: Random Forest
rf_model = RandomForestClassifier()
rf_model.fit(X_train, y_train)
print(f"RF Accuracy: {rf_model.score(X_test, y_test)}")

# Model 2: SVM (immediate comparison in the same notebook)
svm_model = SVC()
svm_model.fit(X_train, y_train)
print(f"SVM Accuracy: {svm_model.score(X_test, y_test)}")

Another major advantage is the ability to share experimental results directly as documents.

# Visualizing learning curves
plt.figure(figsize=(10, 6))
plt.plot(training_scores, label='Training Accuracy')
plt.plot(validation_scores, label='Validation Accuracy')
plt.legend()
plt.title('Model Learning Progress')

4. Educational and Research Applications

Many universities and companies use Jupyter Notebook for the following purposes:

Lecture Materials: Creating teaching materials that integrate code and explanations
Tutorials: Step-by-step learning content
Research Notes: Electronic lab notebooks recording experimental procedures and results
Reports: Creating technical reports including analysis results

Latest Developments in the Jupyter Ecosystem

JupyterLab: Next-Generation Interface

JupyterLab, which had its stable version released in 2018, is an integrated development environment that includes Notebook functionality.

Main Features:

Multi-tab support
File browser
Terminal integration
Extension system
More flexible layout

JupyterHub: Multi-User Environment

Provides an environment where multiple users can use it simultaneously for organizations.

Use Cases:

Simultaneous use in university classes
Corporate analysis team environments
Foundation for cloud service provision

Voilà: Dashboard Creation

A tool that converts Notebooks into web applications.

Application Scenarios:

Interactive reports
Data dashboards
Machine learning model demos

Jupyter AI (2023~)

An extension released in August 2023 that integrates generative AI into Notebooks⁴.

Main Features:

Code explanation and generation
Error correction assistance
Content summarization
Notebook generation from natural language

Corporate Use Cases

Netflix

Adopted as the standard tool for data science workflows. Initially used for analysis, but now also utilized for general data access.

Microsoft

Provides Jupyter as a cloud service through Azure Notebook. Also advancing integration with VS Code, with the Jupyter extension downloaded over 40 million times as of July 2022.

Google

Contributing to the education and research community by providing free Jupyter environment through Colab.

Other Major Companies

IBM, Amazon (SageMaker), many financial institutions, pharmaceutical companies, etc., have adopted it as their data analysis infrastructure.

Cloud services like Google Colab are based on Jupyter, and by using them appropriately according to purpose, more efficient development is possible. For those learning machine learning and data science, Jupyter Notebook has become a new document format that integrates thinking, code, and results, serving as a standard tool in scientific computing. It is expected to continue evolving into an easier-to-use and more powerful tool with further AI integration and cloud collaboration in the future.

Sources

Project Jupyter | Home - Jupyter Official Site
As Project Jupyter celebrates 20 years - UC Berkeley CDSS (Fernando Pérez Interview)
Project Jupyter - Wikipedia - Wikipedia (Project History)
Jupyter AI - Jupyter Official (About AI Features)