Based on comment by Ben, it might be a good idea for community_overlap to return the raw list of pairwise overlaps so that we can give an option for Ostats to return them.
This was originally because he got very different results whether he used mean or median to get the O-stat. We should consider maybe trying different ways of calculating O-stat that are more robust to whether we choose mean or median.
Based on comment by Ben, it might be a good idea for community_overlap to return the raw list of pairwise overlaps so that we can give an option for Ostats to return them.
This was originally because he got very different results whether he used mean or median to get the O-stat. We should consider maybe trying different ways of calculating O-stat that are more robust to whether we choose mean or median.