Right now the HClustering often generates long 'strings' of graphs where the depth of the graph is very deep and each level is only a binary split with just one/few node(s) joining in. This means the visualization is just a bunch of concentric circles. We need to review the find_root/find_parent logic to see if we can generate a flatter tree where sibling counts are higher.
Right now the HClustering often generates long 'strings' of graphs where the depth of the graph is very deep and each level is only a binary split with just one/few node(s) joining in. This means the visualization is just a bunch of concentric circles. We need to review the find_root/find_parent logic to see if we can generate a flatter tree where sibling counts are higher.