v0.8.2

What's Changed

• Task/mean cov pelt cost: Add support for multivariate Gaussian negative log likelihood cost by @johannvk in #25
• Task/weak numba dependency by @johannvk in #26
• Patch: Fix flaky tests of the soft import of numba jit/njit/prange by @johannvk in #31

Full Changelog: v0.8.1...v0.8.2

v0.8.1

What's Changed

• fix: use same code-quality test as sktime by @Tveten in #18
• Set up documentation website by @Tveten in #20
• Store change point indices as integers, not floats. by @johannvk in #21

Full Changelog: v0.8.0...v0.8.1

v0.8.0

What's Changed

• Feature: Score for change in the mean and/or covariance matrix by @johannvk, @Tveten inv #16
• API: Unify naming of scores and costs. Always underscores between separate words. For example: "mean_var" rather than "meanvar", "mean_cov" rather than "meancov" etc.
• Documentation: Improve documentation of score arguments across the library.

New Contributors

• @johannvk made their first contribution in #16

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

v0.7.0

What's Changed

• Inherit from BaseEstimator by @Tveten in #14

Full Changelog: v0.6.0...v0.7.0

Release v0.6.0

Changes

• Update dependencies:

  • sktime >= 0.30. The sktime update is an API break to BaseSeriesAnnotator, which among other things removes the fmt and labels arguments. Since BaseSeriesAnnotator is still experimental, this PR refactors a lot of basic functionality.
  • No upper restriction on numpy and pandas.

• Introduce BaseDetector, which inherits from BaseTransformer in sktime. See here #8 for reasoning. This class is an alternative to BaseSeriesAnnotator as long as it is experimental, but aims to not stray too far away. It also removes the fmt and labels arguments.

• Add base classes per detection type:

  • ChangeDetector
  • PointAnomalyDetector
  • CollectiveAnomalyDetector
  • SubsetCollectiveAnomalyDetector

• Refactor all change and anomaly detectors to depend on a suitable subclass of BaseDetector.

• Convert from black to ruff for formatting.