Jupyter Notebook: Complete Guide to Shortcuts, Magic Commands, and Best Practices

# Install
pip install notebook          # classic Jupyter Notebook
pip install jupyterlab        # JupyterLab (modern interface, recommended)

# Launch
jupyter notebook              # opens at http://localhost:8888
jupyter lab                   # opens JupyterLab
jupyter notebook --no-browser # start without opening browser
jupyter notebook --port 8889  # use a different port

# Place cursor inside the parentheses and press Shift+Tab
import pandas as pd
pd.read_csv(  # Shift+Tab here shows the full signature and docs

pd.read_csv?    # show docstring
pd.read_csv??   # show source code

# Time a single expression
%time sum(range(1_000_000))
# CPU times: user 23.4 ms, sys: 0 ns, total: 23.4 ms

# Average over multiple runs (more accurate)
%timeit sum(range(1_000_000))
# 23.4 ms ± 312 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)

# Time an entire cell
%%timeit
total = 0
for i in range(1_000_000):
    total += i

# Run shell commands with !
!pip install pandas
!ls -la
!git status
!cat requirements.txt

# Capture output into a Python variable
files = !ls *.csv
print(files)  # => ['data.csv', 'test.csv']

# Pass Python variables to shell
filename = "data.csv"
!wc -l {filename}

# Write cell content to a file
%%writefile my_script.py
def hello(name):
    return f"Hello, {name}!"

print(hello("World"))

# Load a file into a cell
%load my_script.py

# Run a Python file
%run my_script.py
%run my_script.py --arg1 value1  # with arguments

# Render matplotlib plots inline (in the notebook)
%matplotlib inline

# Higher resolution plots
%config InlineBackend.figure_format = 'retina'

# Interactive plots (requires ipympl)
%matplotlib widget

# Profile a function
%prun my_function(data)

# Line-by-line profiling (requires line_profiler)
%load_ext line_profiler
%lprun -f my_function my_function(data)

# Memory profiling (requires memory_profiler)
%load_ext memory_profiler
%memit my_function(data)

# List all variables in scope
%who          # brief list
%whos         # detailed table with types and values

# Clear all variables
%reset

# Show command history
%history

# Change working directory
%cd /path/to/directory
%pwd  # print working directory

# Auto-reload modules (great for development)
%load_ext autoreload
%autoreload 2  # reload all modules before executing code

# Display all output (not just last expression)
%config InteractiveShell.ast_node_interactivity = 'all'

from IPython.display import display, HTML, Image, Markdown, JSON

# Render HTML
display(HTML("<h2 style='color:blue'>Hello from HTML</h2>"))

# Render Markdown
display(Markdown("## Section\n\n- Item 1\n- Item 2"))

# Display an image
display(Image(url="https://example.com/image.png"))
display(Image(filename="local_image.png", width=400))

# Display JSON interactively
display(JSON({"key": "value", "nested": {"a": 1}}))

# Multiple outputs from one cell
x = 42
y = "hello"
display(x, y)  # shows both

import pandas as pd

# Show more rows/columns
pd.set_option('display.max_rows', 100)
pd.set_option('display.max_columns', 50)
pd.set_option('display.float_format', '{:.2f}'.format)

df = pd.read_csv('data.csv')

# Styled display
df.head(10).style.highlight_max(color='lightgreen') \
                  .highlight_min(color='lightcoral')

# Progress bar for long operations (requires tqdm)
from tqdm.notebook import tqdm
for row in tqdm(df.itertuples(), total=len(df)):
    pass

# Install useful extensions
pip install jupyterlab-git          # Git integration
pip install jupyterlab-lsp          # Language server (autocomplete, linting)
pip install jupyterlab-code-formatter  # Code formatting
pip install ipywidgets              # Interactive widgets

# Enable widgets
jupyter nbextension enable --py widgetsnbextension

import ipywidgets as widgets
from IPython.display import display

# Slider
slider = widgets.IntSlider(value=5, min=0, max=20, description='Value:')
display(slider)

# Interactive function
@widgets.interact(x=(0, 10), y=(0, 10))
def multiply(x=5, y=5):
    print(f"{x} × {y} = {x * y}")

# Always verify your notebook runs top-to-bottom cleanly
Kernel → Restart & Run All

# Bad: one giant cell
import pandas as pd
import numpy as np
df = pd.read_csv('data.csv')
df = df.dropna()
df['new_col'] = df['a'] * df['b']
result = df.groupby('category').mean()
result.plot()

# Good: separate logical steps into cells
# Cell 1: imports
# Cell 2: load data
# Cell 3: clean data
# Cell 4: feature engineering
# Cell 5: analysis
# Cell 6: visualization

## Data Loading

Load the raw dataset from the CSV file. The dataset contains 10,000 rows
with the following columns: `id`, `date`, `value`, `category`.

**Note:** Missing values in `value` are dropped in the next step.

# Export to various formats
jupyter nbconvert notebook.ipynb --to html
jupyter nbconvert notebook.ipynb --to pdf
jupyter nbconvert notebook.ipynb --to script  # converts to .py
jupyter nbconvert notebook.ipynb --to markdown