# Changelog

## Score

- CAI 54 → 57 (+2.8)
- Rubric changed (rubric-2026.10.5 → rubric-2026.11.2) — scores are not directly comparable.

## Lenses

- Code Health 73 → 73 (+0.0)
- Architecture 88 → 88 (+0.0)
- Maturity 38 → 46 (+7.6)
- Readiness 58 → 57 (-1.0)
- Security 80 → 73 (-7.1)

## Resolved (4)

- Inconsistent naming for cluster centers/prototypes. GeneralizedKMeansModel uses a method clusterCenters() to retrieve centers, while KMedoidsModel uses properties medoids and medoidIndices. 'Centers' and 'Medoids' are semantically distinct (mean vs. actual data point), but the API access pattern differs (method vs. property). This inconsistency in access patterns (method vs property) for similar conceptual outputs (the representative points of clusters) is confusing.
- Outdated: org.scala-lang.modules:scala-parallel-collections_2.13
- Outdated: org.scalactic:scalactic_2.13
- Redundant setter methods alongside setParams. The class exposes both granular setters (e.g., setK, setDivergence) and a bulk setParams method that accepts all parameters. While granular setters are common in PySpark-style APIs, the presence of both creates API surface bloat. More critically, the bulk setParams signature in the constructor/init does not match the individual setters for all fields (e.g., setMaxIter, setSeed, setTol, setFeaturesCol, setPredictionCol, setWeightCol are present as individual setters but are NOT in the GeneralizedKMeansParams field list or the setParams signature shown, implying they are inherited or mixed in from a base Transformer/Estimator interface, making the distinction between 'params' and 'transformer options' unclear).

## New (1)

- Dependency hygiene PARTLY measured — sbt declarations read, dependency currency not (Maven Central unreachable)

## Architecture

- Unchanged — 0 containers · 1 contexts · 0 edges
