Bukka documentation

Bukka is a CLI for creating a machine learning project scaffold from a dataset or a config file. It handles the repetitive setup work so you can start with code that is already organized.

Contents:

• Getting Started
  • Installation
  • Creating Your First Project
  • Example: Binary Classification
  • Example: Regression Project
  • Using the Project Class
  • Project Structure
  • Next Steps
• Usage Examples
  • Basic Project Creation
  • Common project shapes
  • Tuning the scaffold
  • Next steps
• Configuration
  • Configuration File Format
  • Configuration Sections
  • Command-line arguments
  • Configuration examples
  • Configuration Validation
  • Environment Variables
  • Best Practices
  • Troubleshooting
  • Next Steps
• API Reference
  • Core modules
  • Data management
  • Coding utilities
  • Utilities
  • Environment management
  • Preprocessing

Quick start

Install Bukka:

pip install bukka

Create a new project:

python -m bukka run --name titanic --dataset titanic.csv --target Survived

What the command does:

1. Creates the project folder.

2. Sets up a virtual environment unless you pass --skip-venv.

3. Writes the initial configuration and dependency files.

4. Copies the dataset into the new project.

5. Splits the dataset into train and test sets.

6. Writes starter files such as the data reader and notebook.

7. Adds optional helpers when you pass flags like --mlflow, --dummy, or --tpot.

For the command reference and the config file format, see the other pages in this guide.

Indices and tables

• Index

• Module Index

• Search Page