Mahout
  1. Mahout
  2. MAHOUT-284

In Fuzzy Kmeans, when the distance between centroid and the given point is zero, then it should belong to that cluster with probability 1 and rest with probability zero

    Details

    • Type: Bug Bug
    • Status: Closed
    • Priority: Minor Minor
    • Resolution: Won't Fix
    • Affects Version/s: None
    • Fix Version/s: None
    • Component/s: Clustering
    • Labels:
      None

      Description

      In Fuzzy Kmeans, when the distance between centroid and the given point is zero, then the point should belong to that cluster with probability 1 and rest with probability zero. However, right now, we are not doing that.

      1. Mahout-284.patch
        8 kB
        Pallavi Palleti

        Activity

        Hide
        Pallavi Palleti added a comment -

        This patch fix the issue

        Show
        Pallavi Palleti added a comment - This patch fix the issue
        Hide
        Jeff Eastman added a comment -

        This patch is obsolete and after a couple of hours trying to apply it manually have given up. It has a lot of special case tests for strongCluster and corresponding branching to avoid dividing by zero in the computeProbWeight calculation (which currently has the only test for distance == 0, substituting 1e-10). I wonder about the importance of fixing this in general, but would like to keep looking for a more elegant solution that doesn't involve so much special case testing.

        Show
        Jeff Eastman added a comment - This patch is obsolete and after a couple of hours trying to apply it manually have given up. It has a lot of special case tests for strongCluster and corresponding branching to avoid dividing by zero in the computeProbWeight calculation (which currently has the only test for distance == 0, substituting 1e-10). I wonder about the importance of fixing this in general, but would like to keep looking for a more elegant solution that doesn't involve so much special case testing.
        Hide
        Sean Owen added a comment -

        Sounds like we can't proceed on this particular issue. Perhaps Pallavi can at some point resubmit the patch. I believe there was discussion of fixing some issue like this recently anyway?

        Show
        Sean Owen added a comment - Sounds like we can't proceed on this particular issue. Perhaps Pallavi can at some point resubmit the patch. I believe there was discussion of fixing some issue like this recently anyway?

          People

          • Assignee:
            Unassigned
            Reporter:
            Pallavi Palleti
          • Votes:
            0 Vote for this issue
            Watchers:
            0 Start watching this issue

            Dates

            • Created:
              Updated:
              Resolved:

              Development