Session 01: Modern Data Science Computing Environment Setup

A comprehensive guide to setting up a Python 3.10+ computing environment, virtual environment isolation, package managers (pip, conda), JupyterLab, VS Code, and Google Colab with GPU acceleration.

Module Author: Mahendar Dwi Payana
Syllabus Reference: DTS Kominfo - STEI ITB (Ir. Windy Gambetta, MBA) & SKKNI No. 299 of 2020


1. Why an Isolated Computing Environment Matters

In a professional Data Science project lifecycle, one of the most fatal problems is dependency hell. Machine learning models often depend on specific versions of libraries such as numpy, scipy, pandas, and scikit-learn.

Using a global installation without isolation causes:

  1. Reproducibility Failures: A project that runs on your local machine cannot run on a teammate’s server or cloud machine.
  2. Library Version Conflicts: Upgrading a library for Project B can break the code in Project A.
  3. Operating System Pollution: Piling Python libraries directly into OS system directories can disrupt internal OS utilities.

2. Development Environment Taxonomy

Broadly, data science practitioners combine two types of environments:

Data Science Computing Environment
├── Local Environment (Offline, Full Control)
│   ├── CLI & Package Manager: Python 3.10+, pip, venv, Conda / Mamba
│   └── IDE / Notebook Editor: JupyterLab, Jupyter Notebook, VS Code + Python Extension
└── Cloud Environment (Zero-Config, Hardware Acceleration)
    ├── Google Colaboratory (Free T4/V100 GPU & TPU)
    ├── Kaggle Notebooks (30 hours/week of P100/T4 GPU acceleration)
    └── Cloud Managed Notebooks (AWS SageMaker, GCP Vertex AI Workbench)

3. Step-by-Step Local Environment Setup

Ensure Python 3.10+ Is Installed

Download the official stable release from python.org. Make sure the "Add Python to PATH" option is checked during Windows installation.

python --version

Create & Activate a Virtual Environment (venv)

Use the built-in venv module to create a standalone environment without polluting your main system:

mkdir -p ~/projects/learn-ai-datascience
cd ~/projects/learn-ai-datascience
python3 -m venv env-ds
source env-ds/bin/activate

Install Core Data Science Libraries

Install the verified core data science packages and register the kernel with Jupyter:

pip install --upgrade pip
pip install numpy scipy pandas matplotlib seaborn scikit-learn jupyterlab ipykernel
python -m ipykernel install --user --name=env-ds --display-name="Python (Learn-AI Data Science)"

4. Working with Google Colaboratory (Cloud GPU)

Google Colaboratory is a browser-based Jupyter environment connected directly to Google Cloud Platform.

Key Google Colab Features:

  • Zero-Setup: Popular data science libraries are already pre-installed.
  • Hardware Acceleration: Free access to Nvidia T4 / V100 GPUs and Tensor Processing Units (TPU).
  • Google Drive Integration: Save datasets and model weights directly to cloud storage.

Verifying GPU Acceleration in Colab

Run the following snippet in the first cell of a Google Colab notebook:

import torch
import sys

print(f"Python version: {sys.version.split()[0]}")

if torch.cuda.is_available():
    print(f"GPU acceleration active: {torch.cuda.get_device_name(0)}")
    print(f"GPU memory allocation: {torch.cuda.get_device_properties(0).total_memory / 1e9:.2f} GB")
else:
    print("Running on CPU runtime. To enable GPU: Edit -> Notebook settings -> Hardware accelerator -> T4 GPU.")

5. Full Library Verification Script

Create a file named verify_env.py in your project directory and run it to confirm that all libraries are installed with the correct versions:

"""
Modern Data Science Environment Verification Script
Author: Mahendar Dwi Payana
"""

import sys

def verify_ecosystem():
    packages = [
        "numpy",
        "scipy",
        "pandas",
        "matplotlib",
        "seaborn",
        "sklearn",
    ]

    print("=" * 60)
    print("DATA SCIENCE ENVIRONMENT CHECK RESULT")
    print(f"Python Interpreter Version: {sys.version.split()[0]}")
    print("=" * 60)

    success = True
    for pkg in packages:
        try:
            mod = __import__(pkg)
            version = getattr(mod, "__version__", "Installed (version not detected)")
            print(f"[OK] {pkg:<15} : {version}")
        except ImportError as e:
            print(f"[FAIL] {pkg:<15} : Not installed! ({e})")
            success = False

    print("=" * 60)
    if success:
        print("Congratulations! The entire Data Science ecosystem is ready to use.")
    else:
        print("Attention: Please install the failed packages before continuing.")
    print("=" * 60)

if __name__ == "__main__":
    verify_ecosystem()

6. Self-Practice & Evaluation

  1. Create a new Jupyter notebook named 01_setup_verification.ipynb.
  2. Write a markdown cell documenting your computer’s hardware specifications (RAM, Processor, OS).
  3. Run the verification code above and take a screenshot as proof of completing the initial learning prerequisites.

On this page