Dismiss Notice
You must be a registered member in order to post messages and view/download attached files in this forum.
Click here to register.

Apply control limits to a non-normal distribution

Discussion in 'SPC - Statistical Process Control' started by Nuno Sardo, Nov 9, 2018.

  1. Nuno Sardo

    Nuno Sardo New Member

    Joined:
    Nov 9, 2018
    Messages:
    1
    Likes Received:
    0
    Trophy Points:
    1
    Hello

    I have a process parameter that is controlled this way:

    If the equipment signal response is below a certain level, we know for sure that the result will be lower than 10% and thus reported as <10.

    However, if a signal response is a little bit higher, then I have to actually quantify the value. In this cases, I usually get results around 5-6, which are still below the 10.

    My data distbitution is something like this:

    Sample size - 126
    Analysis result | Number of events | Cumulative Count | Cumulative %
    <10 | 97 | 97 | 77%
    5 | 16 | 113 | 90%
    6 | 9 | 122 | 97%
    7 | 3 | 125 | 99%
    8 | 1 | 126 | 100%

    What would be the best distribution to fit this data? Exponential? Weibull? Gamma?

    The purpose is to define an upper control limit, as the ones used for normal distributions in SixSigma (average+3sigma). How should I proceed in this case?

    Thanks in advance for your support
     
  2. Bob Doering

    Bob Doering Member

    Joined:
    Jul 30, 2015
    Messages:
    37
    Likes Received:
    24
    Trophy Points:
    7
    For the sake of a capability study, you need to quantify all your points - even though the data may (or may not be) handy for ongoing control. Then you can establish an accurate model of your data and calculate the control limits based on that model. It may mean transformation - it depends on the distribution model you end up with.
     
  3. Bev D

    Bev D Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    370
    Likes Received:
    376
    Trophy Points:
    62
    Location:
    Maine
    What are you trying to do? Why do you need an ‘upper limit’?
     

Share This Page