Evo-Learn

AutoML toolkit built on TPOT

Installation • Quick Start • Configuration

---

Features

• Automated Model Selection: TPOT-powered search for classification & regression
• Production Pipeline: Preprocessing with imputation, encoding, and scaling
• Smart Fallbacks: Baseline models when TPOT fails or times out
• Config-Driven: YAML configuration for reproducible runs
• Data Validation: Catches NaN, single-class, and imbalance issues
• Visualizations: ROC/PR curves, residuals, feature distributions
• Type-Safe: Complete type hints throughout
• Secure: Joblib serialization, path traversal protection

Installation

From Source

git clone https://github.com/TurboRx/Evo-Learn.git
cd Evo-Learn
python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
python check_installation.py

Docker

docker-compose up evo-learn
docker-compose run evo-learn-test  # Run tests

Quick Start

Python API

from core import run_automl

result = run_automl(
    data_path="data.csv",
    target_column="target",
    task="classification",
    generations=5,
    population_size=20
)

print(f"Accuracy: {result['metrics']['accuracy']:.4f}")
print(f"Model saved: {result['model_path']}")

Command Line

python cli.py train data.csv target --task classification --generations 5
python cli.py predict model.pkl test_data.csv --target target
python cli.py evaluate model.pkl test_data.csv target

Configuration

Create config.yaml:

handle_categoricals: true
impute_strategy: median
scale_numeric: true
output_dir: models
n_jobs: -1

Use with:

result = run_automl(
    data_path="data.csv",
    target_column="target",
    config_path="config.yaml"
)

Project Structure

.
├── core.py              # Main AutoML logic
├── preprocessing.py     # Data preprocessing
├── visualization.py     # Plots and dashboards
├── cli.py              # Command-line interface
├── utils.py            # Utility functions
├── examples/           # Usage examples
└── tests/             # Test suite

Examples

See examples/ directory:

• 01_basic_classification.py
• 02_regression_with_missing_data.py
• 03_imbalanced_classification.py
• 04_categorical_features.py
• 05_new_features_demo.py

Testing

pytest tests/
pytest tests/ --cov=. --cov-report=html

Development

pip install -e .
pre-commit install
black .
flake8 .
mypy .

License

MIT License - see LICENSE file

Contributing

Pull requests welcome. Please ensure:

• Tests pass
• Code formatted with Black
• Type hints included
• Documentation updated

---

Note: Requires Python 3.10+