1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.
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.

DoE with a large number of categorical variables

Discussion in 'DOE - Design of Experiments' started by n00bster, Jan 12, 2023.

  1. n00bster

    n00bster New Member

    Joined:
    Jan 12, 2023
    Messages:
    3
    Likes Received:
    0
    Trophy Points:
    1
    Hi All,

    I have been asked to design a DoE experiment for a project involving food supplement formulations. However, I am not sure DoE can even be used for this study so am asking for some advice.

    We have 20 active ingredients that can be mixed in different quantities and combinations to produce a formulation. A formulation is successful based upon non-quantitative measurements (looks, taste, smell etc).

    Based on this information, I believe this means that the study would use 20 non-quantitative inputs (addition of an ingredient) and a handful of non-quantitative outputs (looks: pass/fail, taste: pass/fail etc). Is it even possible to do DoE based on these inputs and outputs? I was under the impression that DoE relies upon statistically analysing the effects of changing multiple quantitative inputs at once (as opposed to a one-by-one approach).

    If anyone can advise as to whether an approach like this is possible I would be most appreciative.

    Thank you
     
    n00bster, Jan 12, 2023
    #1
  2. Miner

    Miner Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    577
    Likes Received:
    493
    Trophy Points:
    62
    Location:
    Greater Milwaukee USA
    You seem to be describing an OFAT (One Factor At a Time) experiment. OFATs are very inefficient, requiring many more experiments than necessary and cannot identify potential interactions. Results are often inconclusive.

    Since you are experimenting with a formulation, you should use a mixture design DOE for the input factors. Another issue is having attribute outputs as this will drive the need for a very large number of replicates.
     
    Miner, Jan 12, 2023
    #2
    Andy Nichols likes this.
  3. n00bster

    n00bster New Member

    Joined:
    Jan 12, 2023
    Messages:
    3
    Likes Received:
    0
    Trophy Points:
    1
    Thank you for your response. I have looked through the link you posted about mixture design DOE and it seems very different from what I understand of DOE design. Do you have any further links to reading that could help me to understand this further?
     
    n00bster, Jan 12, 2023
    #3
  4. Bev D

    Bev D Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    606
    Likes Received:
    664
    Trophy Points:
    92
    Location:
    Maine
    Please understand that DoE (Design of Experiments) is a name for a very large, broad and flexible set of experimental study designs. Some people like to think and say that DoE only covers a few narrow types of studies. They are wrong.
     
    Bev D, Jan 12, 2023
    #4
    Andy Nichols likes this.
  5. Miner

    Miner Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    577
    Likes Received:
    493
    Trophy Points:
    62
    Location:
    Greater Milwaukee USA
    Do you have specific analysis software that you plan to use? Some software have good help sections. See Minitab's for example.

    This article is more in-depth. You can search for more using "mixture experiments" as the keywords.
     
    Miner, Jan 12, 2023
    #5
    Andy Nichols likes this.
  6. n00bster

    n00bster New Member

    Joined:
    Jan 12, 2023
    Messages:
    3
    Likes Received:
    0
    Trophy Points:
    1
    Thank you all, I will indeed be using Minitab (which I have never used before).
    Looks like I have plenty of reading to do :)
     
    n00bster, Jan 12, 2023
    #6
  7. Miner

    Miner Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    577
    Likes Received:
    493
    Trophy Points:
    62
    Location:
    Greater Milwaukee USA
    Good luck. While I have decades of hands-on experience with DOEs (fractional, full, response surface and Taguchi), my knowledge of mixture DOEs is purely theoretical.
     
    Miner, Jan 12, 2023
    #7
    n00bster likes this.