v0.8.0

What's Changed

• Easier configuration of k-means by @fnattino in #97

Full Changelog: v0.7.0...v0.8.0

v0.7.0

What's Changed

• remove an ISSN (in capital letters) by @hugoledoux in #87
• Joss paper by @fnattino in #88
• Update Python versions by @fnattino in #91
• Improve k-means by @fnattino in #92
• General fixes by @fnattino in #93
• Allow to JSON serialize array-like objects by @fnattino in #94

Full Changelog: v0.6.2...v0.7.0

v0.6.2

Bug fix in tests, improvements in documentation, JOSS manuscript included (for detailed changes see CHANGELOG).

v0.6.1

fix release files

v0.6.0

This release introduces few improvements in the co- and tri-clustering algorithms, as well as in the way in which refined-cluster averages are computed - full description is available in the CHANGELOG.

v0.5.0

This release introduces:

• a k-means implementation for tri-clustering;
• a utility function to calculate cluster-based averages for tri-clustering;
• the best k value in k-means is now selected automatically using the Silhouette score.

v0.4.0

This release mainly includes:

• a new tri-clustering algorithm (with both a Numpy- and a Dask-based implementation);
• some utility functions to estimate the memory requirement of a co-clustering analysis and to calculate co-cluster averages.

For all changes, see CHANGELOG.rst.

v0.3.0

Merge pull request #31 from phenology/development

Development

v0.2.1

With this release we solve some dependency issues that made installation with pip fail

v0.2.0

This release mainly introduces:

• some improvements in the Dask-based co-clustering implementation;
• a low-memory version of the Numpy-based co-clustering implementation (with Numba acceleration).

For a more detailed list of changes, have a look at CHANGELOG.rst.