I think it can make sense. However, since I've realized some problems with what we (or at least I :-) said yesterday, and want to make sure we are all on the same page, I'd like to go into a bit more detail. The idea would be something like the following?

class KMeansLloydsClustering: public PartitionalClustering { KMeansLloydsClustering(EmbeddingAdaptor embed, unsigned int k, unsigned int num_iterations); void refine(unsigned int num_iterations); };

so you would do: clu= KMeansLloydsClustering([[0,1], [1,2]], 2, 10) print clu.get_cluster(0), clu.get_cluster(1) clu.refine(10)

Whatever is done should also have similar things done on the Metric side.

BTW, I split up various headers, so make sure you update before you dive in.

On Thu, Jun 14, 2012 at 12:26 PM, Barak Raveh barak.raveh@gmail.com wrote:

> Hi, > Daniel and I discussed a little bit consolidation of clustering things in > statistics. I am not touching anything out of the statistics / kmeans > module for now, but please tell me if it is agreed that things will work as > follows: > * The Embedding family of classes (used to embed data in vector form) will > remain as is > * There will be a "Clustering" class from which all clustering algorithms > will derive, with: > * constructor that takes either Embedding class, or EmbeddingAdaptor for > implicit conversions from e.g., vector form > > > -- > Barak > > _______________________________________________ > IMP-dev mailing list > IMP-dev@salilab.org > https://salilab.org/mailman/listinfo/imp-dev > >