FEDOT is an open-source AutoML framework designed for automated machine learning pipeline generation. It targets data scientists and researchers seeking to automate the process of building, optimizing, and deploying ML models for classification, regression, and time series forecasting tasks. FEDOT leverages evolutionary algorithms to design complex, composite pipelines, offering flexibility, extensibility, and integrability with popular ML libraries.

How It Works

FEDOT employs a generative design approach based on evolutionary algorithms to construct machine learning pipelines. These pipelines are represented as graphs, enabling the management of complex interactions between data preprocessing steps and modeling blocks. This evolutionary core allows for the exploration of diverse pipeline structures and hyperparameters, aiming to discover optimal solutions for various data types and tasks.

Quick Start & Requirements

• Install: pip install fedot or pip install fedot[extra] for image/text/DNN support.
• Requirements: Python 3.7+. Docker images are available.
• Docs: FEDOT Documentation
• Examples: FEDOT Examples Repository

Highlighted Details

• Supports classification, regression, and time series forecasting.
• Integrates with Scikit-learn, CatBoost, XGBoost, and custom libraries.
• Offers various hyperparameter tuning methods.
• Pipelines can be exported to JSON for reproducibility.

Maintenance & Community

• Supported by the Natural Systems Simulation Lab at ITMO University.
• Active development on LLM integration, meta-learning, and optimization core.
• Community: Telegram Channel
• Roadmap: Implied through ongoing research and development tasks.

Licensing & Compatibility

• License: 3-Clause BSD.
• Compatibility: Permissive license suitable for commercial use and integration with closed-source projects.

Limitations & Caveats

The framework is under active development, with ongoing research into advanced features like LLM integration and meta-learning. Specific performance benchmarks or detailed comparisons against other AutoML tools are not explicitly detailed in the README.