iSixSigma

Yiled vs. Sigma

Six Sigma – iSixSigma Forums Old Forums General Yiled vs. Sigma

Viewing 3 posts - 1 through 3 (of 3 total)
  • Author
    Posts
  • #51213

    newbie
    Participant

    Forum,
    Can a case be made to simply use a yield based metric on historical process data to baseline process performance?  It appears to me as if Z/sigma scores are yield-based anyway and often confusing and highly subjective (and potentially inaccurate)….why invoke the complexity of these metrics if a simple yield metric can be obtained from actual process data?  Thanks!

    0
    #177245

    John Knauf
    Participant

    In my opinion, an actual yield based on historical data should always take precedence over a calculated estimate, if for no other reason than to keep the real “state of nature” in the process evaluation.  Estimates alone, if used sequentially over time, will be affected by the same bedevilment that affects every process – the probability function is multiplicative, causing an ever growing delta between the process model and its reality.  I’ve seen this problem in a number of companies. 
    However, many Black Belt practitioners will tell you that estimates are often all you have to work with.  This happens because the process in question may have unreliable historical data, or no data at all.  When the first happens you can build a model only after a serious data purge, and even then you many find the process dispersion suspiciously large.  When the second happens you have to make estimates based on experimental or experiencial information.  Either case requires careful monitoring to test for reliability. 
    But getting back to the “simple yield metric” question – I would always start with that because, in my opinion at least, it’s always a bad idea to lose contact with your data.   

    0
    #177246

    newbie
    Participant

    Thanks so much!

    0
Viewing 3 posts - 1 through 3 (of 3 total)

The forum ‘General’ is closed to new topics and replies.