Uploaded image for project: 'Architecture for Learning Enabled Correlation (ALEC)'
  1. Architecture for Learning Enabled Correlation (ALEC)
  2. ALEC-107

Hellinger Distance for ALEC: Using Key Customer Data and Compare results to Basic DBSCAN

    XMLWordPrintable

Details

    • Story
    • Status: Resolved (View Workflow)
    • Minor
    • Resolution: Fixed
    • None
    • None
    • None

    Description

      We compare the performance of the DBSCAN clustering engine using "AlarmInSpaceAndTimeDistance" vs DBSCAN using Hellinger Distance. This includes cycles to test multiple combinations of said distance measures.

      Eg.: DBSCAN only using Hellinger Distance with different epsilon values (tuning). Or, modify "AlarmInSpaceAndTimeDistance" formula to include Hellinger Distance.

      We assume that Hellinger Distance has been implemented in ALEC and can be used/modified.

      DONE CRITERIA:

      • Train Neural Network to learn time distribution with key customer data.
      • Compare DBSCAN+AlarmInSpaceAndTimeDistance with DBSCAN+Hellinger Distance using key customer data.
      • Make an assessment of Hellinger Distance impact to ALEC existing correlation methods.
      • Research write-up (confluence)

       
      This doesn't include: * Hellinger Distance implementation in ALEC.

      Attachments

        Issue Links

          Activity

            People

              gmantecon Gerardo Mantecon
              joseanes Jose
              Votes:
              0 Vote for this issue
              Watchers:
              2 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved: