Home › Metrics › Clustering metrics

Homogeneity

Not corrected for chance, and not symmetric. Giving every sample its own cluster scores 1, which is why this number alone never justifies a clustering — AdjustedRand is what sees through that case.

Members

Member	What it does
Homogeneity.Score	Whether each cluster holds samples of a single class.