ML Training Practice (uv)

Package-first machine learning training repository using uv.

Quick start

1. Sync dependencies:

   uv sync

2. Run tests:

   uv run pytest

3. Run lint:

   uv run ruff check .

4. Start JupyterLab:

   uv run jupyter lab

Project structure

src/ml_training_practice/  # Reusable training code
tests/                     # Unit and smoke tests
notebooks/                 # Experiment notebooks
scripts/linear_regression/ # Linear regression practice scripts
scripts/logistic_regression/ # Logistic regression practice scripts
data/raw/                  # Local raw datasets (gitignored)
data/processed/            # Local processed datasets (gitignored)
models/                    # Saved model artifacts (gitignored)
reports/figures/           # Generated figures (gitignored)

Typical workflow

• Build reusable logic in src/ml_training_practice.
• Use notebooks for experiments, but import package code instead of duplicating logic.
• Keep large/local artifacts in data/, models/, and reports/figures/.

Make shortcuts

Run common tasks with short commands:

make sync
make lint
make test
make train
make notebook

Polynomial Regression Practice

Run the polynomial degree comparison (default: degrees 1..5):

uv run python scripts/linear_regression/train_housing_polynomial.py

Or via Make:

make train-poly