bukka 0.0.3

ML Project Quickstart

📖 bukka: Django-Inspired ML Infrastructure CLI

bukka is a Python command-line interface (CLI) tool designed to dramatically reduce the boilerplate and setup time for new Machine Learning (ML) projects. Inspired by the structure and speed of the Django framework's startproject command, bukka lets you instantly scaffold a robust, standardized, and ready-to-use project infrastructure.

---

✨ Features

• Django-Inspired Structure: Creates a logical, maintainable folder hierarchy optimized for ML workflows (data, models, notebooks, scripts).
• Automated Environment Setup: Automatically generates a Python virtual environment (.venv) to isolate your project dependencies.
• Dependency Management: Creates a starting requirements.txt file with essential ML packages (e.g., NumPy, Pandas, Scikit-learn).
• CLI Simplicity: Use simple, intuitive commands to create a complete project skeleton in seconds.

---

🚀 Quick Start

1. Installation

bukka is available on PyPI.

pip install bukka

2. Creating a New Project

Use bukka command, similar to Django, followed by your desired project name.

# Example: Create a new project named 'titanic'
python -m bukka -n titanic -d titanic.csv

This command will:

1. Create the folder in the root directort: titanic/
2. Create and configure a virtual environment: titanic/.venv/
3. Generate the initial dependency file: titanic/requirements.txt
4. Install the packages in the requirements.txt.
5. Copy the data file to your data folder.
6. Split the dataset into a training and test set.
7. Provides a reasonable pipelines based on an adhoc scan of your dataset.

(Coming soon, the command will also do the following):

8. Provide a few baseline models you can compare to, e.g. using random guessing.
9. Provide placeholder utility classes you can customize for your project.
10. Initialize MLFlow to track your parameters and results.
11. Provide starter notebooks, so you can get to machine learning ASAP.

🌳 Standard Project Structure

When you run python -m bukka.bukka -n <name>, the following standardized structure is created, ensuring consistency across all your ML projects:

<project_name>/
├── .venv/                         # Isolated Python Virtual Environment
├── data/                          # Storage for raw, processed, and external data
│   ├── test/                      # Unprocessed, immutable source data
│   ├── train/                     # Cleaned and processed data ready for modeling
├── pipelines/                     # Pipelines
│   ├── __init__.py                # Makes 'pipelines' a Python package
│   ├── baseline/                  # Placeholder, this will store pipelines that provide baselines (e.g. naive classifiers)
│   ├── candidate/                 # Contender pipelines
│   ├── generated/                 # Placeholder, this will contain pipelines generated by the schema analyzer.
├── scripts/                       # Python scripts for automation (currently empty)
├── requirements.txt               # Project dependencies file

---

🎶 About the Name

The library is named after Bukka White, a country blues guitarist whorecorded from the late 30s and 40s, before being rediscovered in the 60s. One of my favorites by him is Fixin' to Die Blues. Country blues guitarists use their technical skill with the guitar to back their singing to create a very complete sound, using their thumb to play rhythym and their fingers to play a melodyand harmonize over that giving it the sound of multiple guitarists.

The choice to name the library after a guitarist is inspired, like much of the rest of this project, by the library Django, named for Django Reinhardt, another phenomenal guitarist.

---

🤝 Contributing

We welcome contributions! If you have suggestions for new structural templates, essential starter packages, or commands, please open an issue or submit a pull request.

---

📄 License

This project is licensed under the Apache License. See the LICENSE file for details.