Precpecified number of clusters

Issue #35 resolved
created an issue

Is there a way to predefine final number of clusters, something like this

community.partition(nxgraph, number_of_clusters = N) ?

Comments (4)

  1. Anvar reporter

    Seems like it is very convenient to have an opportunity to stop at some prespecified hierarchical level, and k-means is not such an algorithm it doesnt provide hierarchy =(

    upd. sorry my bad, community.generate_dendrogram stored hierarchy

  2. Thomas Aynaud repo owner


    You have several ways to work on number of communities, but none will allow you to specify exactly N.

    1) Exploit the dendogram of generate_dendrogram 2) Build yourself a hierarchy, for instance by applying the algorithm recursively on each community 3) Use the resolution parameter.

    Each will allow to get more communities.


  3. Log in to comment