# Changelog

## 0.9.2 (2026-10-01)

*   [BUGFIX]  Deadwood could refine some subclusters too eagerly.

*   [R] In `deadwood`, the returned object is now equipped with the `cluster`
    attribute.


## 0.9.1 (2026-09-29)

*   [NEW FEATURE]  If the dataset is comprised of well-separated clusters
    of heterogeneous densities, Deadwood attempts to split the dataset
    and refine the outlierness markers.

*   [BACKWARD INCOMPATIBILITY] `kneedle_increasing` now returns
    the index of the last element of a vector if a knee/elbow point could
    not be identified.


## 0.9.0 (2026-02-13)

*   [Python] Initial PyPI release.

*   [R] Initial CRAN release.