7

Could anyone suggest some statistical measures to describe the distribution of a dendrogram? If I have two dendrograms, how could can I quantify their structural differences?

onestop
  • 16,816
  • 2
  • 53
  • 83
Eduardas
  • 2,239
  • 4
  • 23
  • 22
  • 1
    A bit more details could be helpful; how they are obtained, are branches weighted? –  Oct 17 '10 at 11:47
  • Let me give you an example. I wrote an algo to cluster variables. Clustering can be represented by a dendrogram. This dendrogram changes over time (based on time series). The structure of the dendrogram slowly evolves and I am look for a measure to describe it (sort of descriptive statistics for dendrograms). – Eduardas Oct 17 '10 at 11:59

1 Answers1

5

See this SO question: https://stackoverflow.com/questions/2218395/how-do-you-compare-the-similarity-between-two-dendrograms-in-r

Community
  • 1
  • 2
    In R there is the `profdpm` package, and an overview of available indices (incl. Fowlkes & Mallows's $B_k$) to compare HCs is available here, http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.164.6189&rep=rep1&type=pdf. – chl Oct 17 '10 at 16:09