💡 Implementation of New Algorithms and Improved project's powerfulness and safeness
v0.0.2
This release introduces several key improvements and features, including the implementation and validation of the three 'CFS Per Group Nested Tree' and 'LexicoRF', parallelization where possible, and a longitudinal dataset handler. Additionally, the codebase is highly documented and more than 95% of it is tested.
Key Features:
- CFS Per Group Nested Tree and LexicoRF: Implemented and validated these algorithms.
- Parallelization: Applied parallelization for performance improvement wherever possible.
- Longitudinal Dataset Handler: Introduced a handler for easy access to non-longitudinal features, longitudinal features group, etc.
- Longitudinal Pipeline: Developed a pipeline specifically for longitudinal-based algorithms, allowing feature groups to pass onto each step of the pipeline.
- Highly Documented Code: Ensured the codebase is well-documented to facilitate understanding and maintenance.
- Extensive Testing: More than 95% of the codebase is tested.
- Hooks and More Tools: Added hooks and other tools for long-term project usage.
- Improved CFS Per Group Algorithm: Introduced a version two of the algorithm, based on the paper's concept level.
- Updated README: The README has been updated with new information.
v0.0.1 - Initial Release
This release marks the initial setup of the Poetry Python project for Scikit-Longitudinal with one first estimator, featuring robust type-checking and an array of linting tools, including pylint, flake8, pre-commit, black, and isort.
Key Features:
- Setup project with one first estimator
- Highly typed Python code to ensure code quality and maintainability.
- Comprehensive linting tools (pylint, flake8, pre-commit, black, isort) integrated into the project to enforce coding standards and consistency.
Estimators:
- Correlation-based Feature Selection (CFS) algorithm: This release includes a refined version of an open-source CFS algorithm, featuring improved typing, testing, runtime optimisation and brand new search algorithms.
- CFS per Group for Longitudinal Data: This release also introduces a Python implementation of a previously Java-based open-source CFS per Group algorithm, tailored for longitudinal data. The Python implementation now is enhanced with parallelism for better performance, testing and highly typing.
We hope you enjoy using this first release and look forward to your feedback and contributions! Cheers!