Persistable 0.5.3

Expose a filtered graph class for persistence-based clustering independently of persistable's choices.

Persistable 0.5.2

Pin numpy version to be strictly less than 2 for now to address incompatibilities with cython.

Persistable 0.5.1

Bump sklearn dependency version.

Persistable 0.5.0

• Save session feature to export and import parameters set with the UI.
• Small changes to interface: flattening style and inline app.
• Remove jupyter_dash dependency and rely on dash 2.11 instead.

Persistable 0.4.6

Pin cython version to be strictly less than 3 for now to address incompatibilities with cython 3.

Persistable 0.4.5

Fix incompatibility with scikit-learn 1.3.0.

Persistable 0.4.4

Fix incompatibility with latest dash release (2.9.2).

Persistable 0.4.3

Set flattening with fewer clustered points as default. The current implementation of the clustering resulting in more clustered points is also more unstable, so it won't be the default until that is addressed.

Persistable 0.4.2

• Add parameter selection option using a single line and a persistence diagram.

• Add subsampling feature.

• Add non-conservative flattening procedure, which results in many fewer unclustered points.

• Various performance improvements.

Persistable 0.4.0

• Fix issues pointed out by JOSS reviewers.

• Automatically publish to pypi on new release.