Metodebevisregistrering

V-measure

V-measure, introduced by Rosenberg and Hirschberg in 2007, is an external clustering evaluation metric based on the harmonic mean of homogeneity and completeness. It measures whether clusters contain only points from a single true class (homogeneity) and whether all points from a true class are assigned to the same cluster (completeness). Values range from 0 to 1.

Sources recorded, not reviewed

Kilderegistrering

Citater kopieret ordret fra metodens kilderegistrering. Ingen påstandsniveauverifikation er udledt heraf.

V-measure (Homogeneity and Completeness Harmonic Mean)

Taksonomisk metoderegistrering · mcdm / model-evaluation

Rosenberg, A., & Hirschberg, J. (2007). V-measure: A conditional entropy-based external cluster evaluation measure. In Proceedings of the 2007 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning (pp. 410-420). · URL

Åbn fuld metode

Kuraterede påstande

Påstande gemt i bevis-loggen, hver med sin egen vurdering.

Ingen kuraterede påstande endnu

Denne visning opfinder ikke en påstandsvurdering, når loggen ingen har.

Relaterede metoder

Genereret fra metodegrafen og vist som maskinelt foreslåede relationer — ingen bevispåstand er udledt.

Taxonomic bucketAdjusted Rand Indexmachine-suggested · Relational suggestion, not evidence.Same method familyDavies-Bouldin Indexmachine-suggested · Relational suggestion, not evidence.Taxonomic bucketFowlkes-Mallows Indexmachine-suggested · Relational suggestion, not evidence.Taxonomic bucketNormalized Mutual Informationmachine-suggested · Relational suggestion, not evidence.Same method familySilhouette Scoremachine-suggested · Relational suggestion, not evidence.