v1.3.0

In version 1.3.0 of Mellon, the major developments include the introduction of the mellon.model.TimeSensitiveDensityEstimator and the addition of serialization capabilities for the predictive function.

Major Updates

• mellon.model.TimeSensitiveDensityEstimator is now available, extending the suite of estimators in Mellon.
• Serialization capabilities for predictive functions have been added, improving the portability and efficiency of predictive models.

Minor Changes and Improvements

• mellon.model.FunctionEstimator.multi_fit_predict is now deprecated in favor of mellon.model.FunctionEstimator.fit_predict.
• The user namespace has been cleaned up for better organization and to facilitate navigation.
• File organization has been revised, with each estimator now located in a separate file.
• The logging system has been improved to provide more information.
• Testing coverage has been increased, improving the robustness and reliability of the software.
• Documentation has been updated, adopting numpy style docstrings for better readability in the source code.
• The Cholesky based Sparse Gaussian Process now includes a covariance rank test.
• ADVI for density inference has been implemented, enhancing the inference capabilities.
• Input validation has been added extensively across all major user-facing classes and functions, improving the robustness of the software.
• Miscellaneous bug fixes and performance improvements have been made throughout the codebase.