Using Google Colab

Abstract:

What is Google Colab?

Google Colaboratory, or Google Colab , is a free, cloud-based Jupyter notebook environment provided by Google. It allows you to write and execute Python code in your browser —with no setup required and free access to GPUs and TPUs for computation-heavy tasks like deep learning. Think of it as Google Docs, but for Python code.

Colab is widely used in data science, machine learning, and education because it provides an accessible, flexible, and powerful coding environment.

Why Use Google Colab for Data Science?

No Installation Needed: Everything runs in the browser. You don't need to install Python, Jupyter, or any packages locally.
Free GPUs and TPUs: Great for machine learning experiments and prototyping.
Seamless Integration with Google Drive: Your notebooks are stored in the cloud and synced across devices.
Built-in Collaboration: Share notebooks and work with your team like you would with Google Docs.
Pre-installed Libraries: Colab comes with many popular Python packages (like NumPy, pandas, matplotlib, TensorFlow, etc.) ready to go.

Getting Started with Google Colab

Step 1: Access Colab

You can open Colab in two main ways:

Direct link: https://colab.research.google.com
Or open a new notebook directly from Google Drive:
Go to Google Drive
Click + New → More → Google Colaboratory

Tip: If you don't see "Google Colaboratory" in the menu, click + Connect more apps and search for it to install.

Step 2: Create a New Notebook

Once you're in Colab:

Click File → New notebook to create a fresh notebook.
You'll see a familiar notebook interface with code cells and text (Markdown) cells , just like in Jupyter Notebooks.

Understanding the Colab Interface

Code Cells: Where you write and execute Python code.
Text Cells: Written in Markdown; great for explanations, notes, and formatting.
Runtime Menu: Manage your execution environment.
Runtime → Change runtime type allows you to select:
- Python version
- Hardware accelerator (None, GPU, TPU)

Working with Files and Google Drive

Colab can mount your Google Drive , so you can access and save datasets, models, and scripts directly.

Mounting Drive:

from google.colab import drive drive.mount('/content/drive')

After running, you'll be prompted to authorize access. Once mounted, your Drive contents will be available under /content/drive/My Drive/ .

Importing and Using Libraries

Many libraries come pre-installed. You can import and use them as usual:

import numpy as np import pandas as pd import matplotlib.pyplot as plt

If a library isn’t available, you can install it with a !pip command:

!pip install seaborn

Saving Your Work

By default, your notebook is saved in your Google Drive under Colab Notebooks . You can also download it in multiple formats:

File → Download as .ipynb (Jupyter), .py , or .pdf

Collaboration & Sharing

Use the Share button (top right) to invite collaborators.
You can set access permissions (view/comment/edit), just like a Google Doc.

Advanced Tips

Use GitHub Repositories: Open a notebook from GitHub via File → Open notebook → GitHub .
Terminal-like Commands: Use ! before terminal commands:

!ls !pwd !pip install xgboost

Keyboard Shortcuts:
Shift + Enter → Run cell
Ctrl + M + B → Add code cell
Ctrl + M + M → Add text cell

GPU and TPU Support

To use a GPU/TPU:

Go to Runtime → Change runtime type
Select GPU or TPU under "Hardware accelerator"
Check if it's active:

import tensorflow as tf print("GPU:", tf.config.list_physical_devices('GPU'))

Common Pitfalls & How to Avoid Them

Problem	Solution
Notebook disconnects after inactivity	Regularly interact with it, or consider using a local runtime
Session resets and data is lost	Always save important data to Drive
Can't install some libraries	Make sure the package is pip-compatible or use a local runtime