YAAML (Yet Another AutoML)

Lightweight, production-ready AutoML built entirely on scikit-learn with Python 3.12+ modernization.

Quick Start

Installation

# From Test PyPI (current releases)
pip install --index-url https://test.pypi.org/simple/ --extra-index-url https://pypi.org/simple/ yaaml

# From PyPI (when available)
pip install yaaml

# From GitHub (latest development)
pip install git+https://github.com/JordanRex/yaaml.git

# From GitHub releases
pip install https://github.com/JordanRex/yaaml/archive/refs/tags/v0.1.2.tar.gz

Basic Usage

import pandas as pd
from yaaml import YAAMLAutoML
from sklearn.model_selection import train_test_split

# Load your data
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2)

# AutoML in 3 lines
automl = YAAMLAutoML(mode='classification', max_evals=20, feature_engineering=True)
automl.fit(X_train, y_train)
accuracy = automl.score(X_test, y_test)

Key Features

• 🚀 Zero Dependencies: Built entirely on scikit-learn
• 🎯 Multi-Task: Classification and regression
• 🔄 Complete Pipeline: End-to-end ML with intelligent preprocessing
• 🧠 Smart Automation: Feature engineering, selection, hyperparameter optimization
• ⚡ Lightweight: Fast training, minimal resource footprint
• 🔍 Transparent: Full visibility into decisions and transformations

Advanced Usage

# Handle mixed data types automatically
automl = YAAMLAutoML(
    mode='classification',
    imputation_strategy='iterative',  # Advanced missing values
    encoding_method='target',         # Target encoding
    sampling_strategy='balanced',     # Handle imbalanced data
    feature_selection=True,          # Intelligent feature selection
    max_evals=50,
    cv_folds=10
)

# Works with messy real-world data
mixed_data = pd.DataFrame({
    'numeric': [1.5, 2.3, np.nan, 4.1],
    'categorical': ['A', 'B', 'A', 'C'],
    'text': ['good', 'excellent', 'poor', 'good']
})

automl.fit(mixed_data, target)
predictions = automl.predict(new_data)

Project Status

Current: v0.1.2 - Production ready with comprehensive Python 3.12+ modernization

• ✅ Complete: Classification, regression, feature engineering, hyperparameter optimization
• ✅ Tested: Realistic test suite achieving 70% accuracy on challenging datasets
• ✅ Modern: Full Python 3.12+ typing with walrus operator and union types
• ✅ Validated: Enhanced CI/CD pipeline with performance benchmarking

See docs/status.md for detailed roadmap and docs/testing.md for testing infrastructure.

Documentation

• Testing Infrastructure: Comprehensive test suite details
• Project Status: Current status and roadmap
• Contributing: Development guidelines
• Examples: Usage examples and tutorials

License

MIT License - see LICENSE for details.