Lodestar.Metrics 0.2.0. This page is frozen at that release. Read the current documentation for what main says now. A link to a decision or a migration page follows main, and leaves the archive.

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.