fun-ml-projects

Small projects collected over the years.

Python environment (pyproject.toml)

The file pyproject.toml at the repository root pins dependencies for these directories only:

Directory	Notes
finance/	Equities notebook (Yahoo Finance, Plotly, Wikipedia / read_html) and corporate-bond GNN notebook (SciPy, scikit-learn, NetworkX).
miscellaneous/	silly_examples.ipynb (NumPy, Pandas, Matplotlib).
time_series/	ARIMA / ARCH / portfolio notebooks (statsmodels, arch, cvxpy, riskfolio-lib, yfinance, …) and basic_rnn_pytorch.ipynb (PyTorch).

It does not apply to other folders in the repo (for example deep_learning/, kaggle/, __udacity__/, and so on). Those projects may need their own environments or extra packages.

Requirements

• uv — install it using the official install guide.
• Python 3.10 or newer — uv can install and pin a version for you (requires-python in pyproject.toml is >=3.10).

Install

Run everything from the repository root (the directory that contains pyproject.toml).

The [notebook] extra pulls in Jupyter and ipykernel so you can open the .ipynb files. PyTorch is a normal dependency because time_series/basic_rnn_pytorch.ipynb needs it, so the first install can be large and take a while.

# Have uv download a Python that satisfies 3.10+ (adjust 3.12 if you like)
uv python install 3.12

# Create the virtual environment
uv venv --python 3.12

source .venv/bin/activate          # Windows: .venv\Scripts\activate
uv pip install -e ".[notebook]"

If you prefer not to activate the venv, target it explicitly:

uv venv --python 3.12
uv pip install --python .venv/bin/python -e ".[notebook]"

On Windows, use .venv\Scripts\python.exe in place of .venv/bin/python.

uv pip install reads the same pyproject.toml metadata as pip (dependencies and optional extras such as [notebook]).

Optional: named Jupyter kernel

After the environment is installed and activated:

python -m ipykernel install --user --name fun-ml-projects --display-name "Python (fun-ml-projects)"

Then choose Python (fun-ml-projects) when opening notebooks under finance/, miscellaneous/, or time_series/. In Cursor or VS Code you can instead pick the interpreter .venv/bin/python directly.