v0.2.0 (July, 2025)

🎉 Major Release - Dependency Removal & Codebase Modernization

This release represents a substantial modernization of the timecorr package, removing external dependencies, improving code quality, and enhancing documentation.

🔥 Breaking Changes

• Removed brainconn dependency: Implemented core graph theory functions directly (eigenvector centrality, PageRank, node strength) to eliminate external dependency issues
• Cleaned up public API: Removed non-existent functions from API documentation that were causing import errors

✨ New Features

• GitHub Actions CI/CD: Added automated testing on Python 3.8-3.11 for all pushes and pull requests
• Enhanced documentation: All user-facing functions now have comprehensive NumPy-style docstrings with examples
• Direct graph theory implementation: Native implementations of eigenvector_centrality_und(), pagerank_centrality(), and strengths_und()

🛠️ Improvements

• Code formatting: Applied black and isort across entire codebase for consistent style
• Import organization: Cleaned up import statements throughout the project
• Documentation cleanup: Removed build artifacts and redundant files from docs/ directory
• Tutorial enhancements: Fixed and verified all tutorial notebooks and examples
• Test coverage: Maintained 100% test coverage with 132 passing tests

🐛 Bug Fixes

• Fixed kernel parameter errors in Mexican Hat weights (changed 'var' to 'sigma')
• Fixed reshape errors in applications tutorial
• Resolved deprecation warnings by replacing np.math.pi with math.pi
• Fixed PCA dimension errors by limiting to number of samples
• Fixed padding logic for different array dimensions and list inputs

📚 Documentation

• API Documentation: Updated to reflect only existing functions, removed references to non-existent ones
• Enhanced docstrings: Added comprehensive documentation for all major user-facing functions
• Tutorial improvements: Fixed and verified all Jupyter notebooks and Python examples
• README updates: Enhanced with working examples and current API information

🧹 Cleanup

• Removed outdated files: Cleaned up PNG files, temporary scripts, and build artifacts
• Docker removal: Removed outdated Docker setup with broken notebook
• File organization: Streamlined repository structure removing unnecessary files

🔧 Development

• Modern workflow: Transitioned from nose to pytest for testing
• Linting: Applied comprehensive code formatting and style improvements
• CI/CD: Automated testing pipeline for continuous integration

📦 Dependencies

• Removed: brainconn (replaced with native implementations)
• Updated: All dependencies to supported versions
• Simplified: Reduced external dependency footprint

🧪 Testing

• 132 tests passing: Complete test suite with comprehensive coverage
• Cross-platform: Verified on multiple Python versions (3.8-3.11)
• Automated: GitHub Actions ensures tests run on every change

---

See CHANGELOG.md for detailed changes.

v0.1.7 (December, 2023)

This version updates the requirements to enable compatibility with ARM macs (M1, M2, M3).

Full Changelog: v0.1.6...v0.1.7

0.1.6

• updates to specify requirements
• weighted timepoint decoding code updated for timecorr-paper

timecorr

• updated setup.py
• add brainconn as optional dependency

v0.1.0

Initial release of a Python toolbox for discovering and analyzing high-order dynamic correlations in multi-subject multivariate timeseries data.

Readthedocs page: http://timecorr.readthedocs.io