iSixSigma

DOE for Attributre response

Six Sigma – iSixSigma Forums Old Forums General DOE for Attributre response

Viewing 11 posts - 1 through 11 (of 11 total)
  • Author
    Posts
  • #43358

    Mel
    Participant

    Hi Guys,
    We are trying to conduct a DOE for a variable factors but an attibute response. Can anyone help me on this?
    What tool am I going to use? Do I need to convert the attibute response into a variable response?
    Please advise
    Sincerely,
    Mel
     
     

    0
    #137468

    Willis Lee
    Member

    Mel, how many levels of the atttribute response would be, based on your anticipation? I guess the difficulty is how to analyze the data if the attribute level is 3 or even fewer. If there is a way to do the conversion, probably it is better.

    0
    #137470

    Mel
    Participant

    Hi Will,
    Yup, The level of the response is only 2 Pass / Fail.Im gonna try how can I convert this into a variable response.
    Im trying to search  what kind of  DOE will I use in this attribute response.
    Thanks
     
    Mel

    0
    #137473

    rodrigues
    Participant

    Hi
    You would need to do repeats of each run, and calculate the % success ( or fail ). Then you will have a continious response.
    You could also try to set up a full factorial ( several levels at your X’s) and analyze the data via binary regression.
    Depends on how many factors you have, how much money you can spend etc etc.
     

    0
    #137474

    Mel
    Participant

    Hi Maria,
     
    Thanks for the advise. We will try and assess cost of the experiment also.
    Regards,
    Mel
     

    0
    #137475

    Swaggerty
    Participant

    You simply need to convert the attribute data into continuous data.  This can be done using multiple subjective evaluation technique and lattice squares.  Give me your email and i’ll send you a presentation

    0
    #137479

    Mel
    Participant

    Hi George,
     
    Here is my email ad  [email protected]
    Thanks!
    Mel

    0
    #137485

    Yassir
    Member

    Mel
        Another approach is to try using Binary Logistic Regression (BLR) technique instead of a stright DOE.  That will help you evaluate the probaility of passing or failing.  This can only work if you have at least one variable factor (X).
     
    Yassir

    0
    #137486

    Swaggerty
    Participant

    Done

    0
    #137487

    BTDT
    Participant

    george:MSE references are hard to find. It seems similar to finding principal components. I’d appreciate a copy of the pitch.Cheers, BTDT6sigmaguru(at)gmail(dot)com

    0
    #137489

    Swaggerty
    Participant

    ok BTDT it’s done

    0
Viewing 11 posts - 1 through 11 (of 11 total)

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