Requirements

These instructions are not complete and will be updated soon

Setting Up Your Local Environment for Data Science

If you want to run everything on your own laptop, follow the steps below. This guide works for macOS, Windows, and Linux.

1. Install Miniforge (Python + Conda/Mamba)

We’ll use Miniforge, a lightweight Python distribution with the conda-forge channel and the mamba package manager (faster than conda).

Download Miniforge for your operating system:

• https://github.com/conda-forge/miniforge/releases

Run the installer and accept the defaults.

After install, close and reopen your terminal.

Test it:

mamba --version

2. Create a Course Environment

We’ll make a clean environment called ds with all required libraries.

mamba create -n ds python=3.11 jupyterlab ipywidgets \
  numpy jax pytorch torchvision torchaudio cpuonly \
  matplotlib plotly
mamba activate ds

Now you have:

• Python 3.11
• JupyterLab (notebooks)
• NumPy, JAX, PyTorch (math/ML libraries)
• Matplotlib and Plotly (visualization)

3. Running JupyterLab

Start JupyterLab inside the environment:

jupyter lab

It will open in your browser at http://localhost:8888

4. Verifying Your Setup

Inside a notebook, try:

import numpy as np, jax, torch
import matplotlib.pyplot as plt, plotly.express as px

print("NumPy:", np.__version__)
print("JAX:", jax.__version__)
print("PyTorch:", torch.__version__)

plt.plot([0,1,2],[0,1,4])
plt.show()

If all runs without errors, you’re good to go.

Troubleshooting

• Command not found: mamba → Make sure Miniforge is installed and your shell restarted.
• Plotly doesn’t show charts → In JupyterLab make sure the plotly extension is installed:

mamba install -n ds -c conda-forge jupyterlab-plotly

GPU acceleration (optional) → On machines with NVIDIA GPUs, install the CUDA-enabled PyTorch build instead of cpuonly:

mamba install -n ds pytorch torchvision torchaudio pytorch-cuda=12.1 -c pytorch -c nvidia

✅ After this setup, you’ll have everything you need locally for the Data Science class.