MBPLS version 1.0.4

The copy attribute now also correctly applies to python lists, e.g. when using multiple data blocks saved in a list. This prevents hidden mutability of these objects.

MBPLS version 1.0.3

Update of the package to the newest scikit-learn version 0.21.2 that requires a different MRO of the base classes to ensure compatibility with the scikit-learn pipeline. Aside this the scikit-learn copy behaviour of input data was implemented.
Furthermore, the convergence criterion of the NIPALS algorithm was changed to default to the Frobenius-Norm together with the ability to choose other norms.

MBPLS version 1.0.2

Update of the package to the newest scikit-learn version 0.21.2 that requires the MultiOutputMixin base class to ensure compatibility with the scikit-learn pipeline.

Initial release

Initial release after the JOSS review phase.