locked
Minimum Support ignored when filter is applied RRS feed

  • Question

  • When I set the Minimum_Support Algorithm parameter and add a filter on a clustering mining model, analysis services still creates clusters with fewer results than the minimum support arguments specifies. It does this even when there are enough records in the population, after the filter is applied, to create clusters larger than the required minimum support.

    Is there a way to force SSAS to respect the Minimum_Support parameter so long as there are sufficient records in the population?

    Thank you,

    Jason

    Thursday, February 17, 2011 7:24 PM

Answers

  • Hi Jason,

    If you use EM (Expectation Maximization (EM) method, is a soft clustering method), which is the default clustering method, some clusters may have a support value that is lower than the specified value. This is because each case is evaluated for its membership in all possible clusters, and for some clusters there may be only minimal support. So I would suggest you read blow link to check which method is suitable:

    http://msdn.microsoft.com/en-us/library/cc280445.aspx

     

    Hope this helps,

    Raymond


    Raymond Li - MSFT
    • Marked as answer by Jason Bowers Monday, March 7, 2011 7:31 PM
    Thursday, February 24, 2011 5:20 AM