AI & Machine Learning

Jupyter Notebook Complete Guide

Master Jupyter Notebook for data science and machine learning development with Python 3.12, JupyterLab, extensions, kernels, and production deployment on Ubuntu 24.04 LTS.

What you'll learn

Install JupyterLab and configure for data science work
Work with Jupyter cells, kernels, and magic commands
Analyse data with Pandas and NumPy
Create visualisations with Matplotlib and Seaborn
Use JupyterLab extensions for productivity
Deploy JupyterHub for team collaboration
Export notebooks to PDF, HTML, and Python scripts

Installation

Bash

1python3.12 -m venv .venv && source .venv/bin/activate
2
3pip install jupyterlab pandas numpy matplotlib seaborn scikit-learn
4pip install jupyterlab-git jupyterlab-lsp python-lsp-server
5
6# Launch JupyterLab
7jupyter lab --port 8888 --no-browser
8
9# Access at http://localhost:8888/?token=...

Interface & Cells

Python

1# Magic commands
2%timeit sum(range(1000))          # time a single expression
3%%timeit                           # time entire cell
4import numpy as np
5np.random.randn(1000).mean()
6
7%matplotlib inline                 # render plots inline
8%load_ext autoreload               # auto-reload modules
9%autoreload 2
10
11# Shell commands
12!pip install plotly
13!ls -la data/
14
15# Keyboard shortcuts (command mode)
16# A — insert cell above | B — insert cell below
17# DD — delete cell | Z — undo | Shift+Enter — run cell

Data Analysis with Pandas

Python

1import pandas as pd
2import numpy as np
3
4# Load data
5df = pd.read_csv('data/sales.csv', parse_dates=['date'])
6
7# Explore
8print(df.shape, df.dtypes)
9df.describe()
10df.isnull().sum()
11
12# Clean
13df = df.dropna(subset=['revenue'])
14df['revenue'] = df['revenue'].clip(lower=0)
15
16# Aggregate
17monthly = (
18    df.groupby(df['date'].dt.to_period('M'))
19    .agg(total_revenue=('revenue', 'sum'), orders=('id', 'count'))
20    .reset_index()
21)
22
23# Merge
24products = pd.read_csv('data/products.csv')
25df = df.merge(products, on='product_id', how='left')

Visualisation

Python

1import matplotlib.pyplot as plt
2import seaborn as sns
3
4# Style
5plt.style.use('seaborn-v0_8-whitegrid')
6sns.set_palette('husl')
7
8fig, axes = plt.subplots(1, 2, figsize=(14, 5))
9
10# Line chart
11axes[0].plot(monthly['date'].astype(str), monthly['total_revenue'])
12axes[0].set_title('Monthly Revenue')
13axes[0].tick_params(axis='x', rotation=45)
14
15# Distribution
16sns.histplot(df['revenue'], bins=50, kde=True, ax=axes[1])
17axes[1].set_title('Revenue Distribution')
18
19plt.tight_layout()
20plt.savefig('charts/revenue_analysis.png', dpi=150)
21plt.show()

Extensions & Productivity

Bash

1# Useful extensions
2pip install jupyterlab-git            # Git integration
3pip install aquirdturtle_collapsible_headings  # collapsible sections
4pip install jupyterlab_execute_time   # cell execution time
5pip install nbformat black            # auto-format with Black
6
7# nbconvert — export notebooks
8jupyter nbconvert --to html   analysis.ipynb
9jupyter nbconvert --to pdf    analysis.ipynb
10jupyter nbconvert --to script analysis.ipynb   # → analysis.py

Deployment

Bash

1# JupyterHub — multi-user team server
2pip install jupyterhub
3jupyterhub --generate-config
4# Edit jupyterhub_config.py for authentication (PAM, OAuth, LDAP)
5jupyterhub -f jupyterhub_config.py
6
7# Docker deployment
8docker run -p 8888:8888   -v $(pwd)/notebooks:/home/jovyan/work   jupyter/datascience-notebook
9
10# Scheduled notebook execution with papermill
11pip install papermill
12papermill analysis.ipynb output.ipynb   -p date "2025-09-18"   -p threshold 0.8