geewynn / featurologists

Featurologists

Main repository for the MLOps Engineering Lab #2 "Feature Store for ML" by Team #2.

Quickstart

# Create a virtual environemnt:
python -m venv ./venv
source venv/bin/activate

# Setup project:
make setup    # installs package, linters, pre-commit hooks
make lint     # beautifies the staged code
make pytest   # runs tests from ./tests/
make nbtest   # executes notebooks from ./notebooks
make nbclean  # cleans outputs in notebooks from ./notebooks

# Run jupyter locally:
make jupyter

NB: pre-commit hooks installed to your system might be very annoying not allowing you to commit your code. If you're pissed off, please do git commit -m "I'm pissed off" --no-verify and push your changes, we'll figure this out.

Repository format

1. Primary python code:

$ tree featurologists
featurologists
├── __init__.py
└── models
    ├── customer_segmentation
    │   └── __init__.py
    └── __init__.py

If you need to add a pip dependency, please add it to setup.py (section install_requires). Development requirements (linters, jupyters, etc) can be added to requirements-dev.txt.

2. Checking your code in jupyter notebooks:

$ tree notebooks
notebooks
└── models
    ├── country_prediction
    └── customer_segmentation

Please don't keep any important code in jupyter notebooks as it can't be versioned or re-used. It's better to import python code via import featurologists in the notebooks and test it.