Jupyter Notebooks & Development Environment

🎯 Learning Objectives

• Use Jupyter notebooks effectively with keyboard shortcuts and magic commands
• Create interactive visualizations with widgets
• Structure notebooks for reproducibility
• Compare Jupyter with other tools (VSCode, Google Colab)

📖 Core Content

3.1 Jupyter Notebook Best Practices

3.2 Essential Magic Commands

python
Run Inline
# runnable
# %matplotlib inline  # Display plots in notebook
# %run other_script.py  # Run another script
# %time  # Time execution of a statement
# %timeit  # Average execution time
# %who  # List all variables
# %load filename.py  # Load code from file
# %%writefile output.py  # Write cell to file
# %debug  # Enter debug mode after error
import numpy as np
# %time example
arr = np.random.randn(1000)
mean_val = np.mean(arr)
print(f"Mean: {mean_val:.3f}")

3.3 Interactive Widgets

python
Run Inline
# runnable
# Note: ipywidgets may not be available in all runtimes
# import ipywidgets as widgets
# from IPython.display import display
#
# slider = widgets.IntSlider(value=5, min=0, max=10, description='k neighbors:')
# display(slider)
#
# @widgets.interact(k=(1, 20))
# def plot_knn(k=5):
#     print(f"Using k={k} for k-NN classification")

3.4 Development Environment Comparison

📝 Practice Questions

Q1: Why clear all outputs before committing notebooks?

Outputs (images, data tables) can be very large and change frequently. They bloat the repository and create meaningless diffs. The notebook's logic (code + markdown) is what matters for version control. Outputs are regenerated when someone runs the notebook. Q2: What's the advantage of cell-based development?

Cells allow iterative development: run one step, inspect results, modify, run again. This is faster than rerunning an entire script. Cells also serve as documentation boundaries — each cell should do one thing with a markdown explanation. It's like literate programming. Q3: When would you use Google Colab vs Jupyter locally?

Use Colab when: (1) you need GPU/TPU (deep learning with free GPU), (2) collaborating in real-time, (3) no local setup needed. Use Jupyter locally when: (1) working with sensitive data (privacy), (2) large datasets (uploading to Colab is slow), (3) need custom packages or configurations. Join Discord PreviousGit & GitHubNextDocker