Repository navigation
🎉 Highlights
MetaBCI v0.2.0 is a significant update that brings modern package management, modular installation support, and comprehensive CI/CD pipeline improvements.
✨ New Features
-
Modular Installation - Install only what you need, reducing dependency complexity
pip install metabci # Core package only
pip install metabci[brainda] # Datasets and algorithms
pip install metabci[brainflow] # Signal acquisition
pip install metabci[brainstim] # Stimulus presentation
pip install metabci[all] # Full installation -
Cross-dataset Transfer Learning - Added pre-training based cross-dataset transfer learning methods
-
New Transfer Learning Algorithms - Added API for MEKT, LST, SAME algorithms
🔧 Improvements
- Modern Build System - Migrated to pyproject.toml, compliant with PEP 517/518 standards
- Python Version Support - Now supports Python 3.9 - 3.13 (dropped EOL Python 3.8)
- Multiple Installation Methods - Supports pip, conda, and uv package managers
- Code Quality - Fixed PEP8 and mypy errors, improved code standards
- MNE Compatibility - Adapted to the latest MNE package version
- Added mne-bids Dependency - Better support for BIDS format datasets
🐛 Bug Fixes
- Fixed BNCI dataset loading issues
- Fixed data hook related bugs
- Fixed pylsl import method
- Fixed PEP8 errors in paradigm module
🔄 CI/CD
- Added comprehensive GitHub Actions CI workflows
- Automated multi-platform, multi-Python version testing (Linux, macOS, Windows)
- Automated PyPI publishing with Trusted Publisher (OIDC)
- Added Dependabot for automatic dependency updates
- Code quality checks with flake8 and mypy
📦 Dependencies
- Updated psychopy version requirements
- Added mne-bids dependency
- Optimized dependency groups to reduce unnecessary installations
📚 Documentation
- Updated README installation instructions
- Added documentation for multiple installation methods