1. Hello and Welcome to The Quality Forum Online...Continuing in the spirit of People Helping People !
    Dismiss Notice
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.

2^3 DOE Problem

Discussion in 'DOE - Design of Experiments' started by Antisa93, Dec 2, 2016.

  1. Antisa93

    Antisa93 New Member

    Joined:
    Dec 2, 2016
    Messages:
    4
    Likes Received:
    0
    Trophy Points:
    1
    Hello everyone.

    I have a problem with my DOE in Statistica software. I have 3 dependent variables, and one independent.

    I have 8 runs, just like on this photo:
    https://postimg.org/image/g01rw6avn/

    And my model has 3-way interaction. But, when i select 3-way interaction i can't get standard error and p of every coefficient. But, when i select 2-way interaction, i get everything and that's ok:
    https://postimg.org/image/t38x0qcdr/

    But, i can't see this for 3-way interaction model. When i select that, this is all i get:
    https://postimg.org/image/y6vs9micl/

    Where is the problem?
    Thanks!
     
  2. Miner

    Miner Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    264
    Likes Received:
    186
    Trophy Points:
    42
    Location:
    Greater Milwaukee USA
    I cannot access file share sites and cannot see your images, but the usual cause of your problem is a fully saturated model (i.e., 0 df left for the error term). You can correct this by adding a replicate or, if feasible, center points.

    What is the system of interest? In some systems (e.g., electro-mechanical), 3-way interactions are extremely rare and if they exist at all are very weak. Whereas in other systems (e.g., chemical) they are quite common.

    BTW, I believe that you meant to state that you have 3 independent variables (i.e., factors) and 1 dependent variable (i.e., response). Otherwise, an interaction would be impossible.
     
    Antisa93 likes this.
  3. Antisa93

    Antisa93 New Member

    Joined:
    Dec 2, 2016
    Messages:
    4
    Likes Received:
    0
    Trophy Points:
    1
    Don't know what's the problem with images, they are working fine to me.

    Yes, you are correct. My mistake in writing, sorry. System of interest is modelling of force in deep drawing proces. Mechanical engineering/metal forming processes.
     
  4. Miner

    Miner Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    264
    Likes Received:
    186
    Trophy Points:
    42
    Location:
    Greater Milwaukee USA
    Is this a computer model such as finite element analysis? I ask because computer models often do not provide a source of error, which prevent the calculation of the F statistics and p-values. You can still calculate the percent contribution using the Sum of Squares.

    Regarding the images, many companies block access to file share sites for security reasons. If you attach them as JPEGs, all can see them.
     
  5. Antisa93

    Antisa93 New Member

    Joined:
    Dec 2, 2016
    Messages:
    4
    Likes Received:
    0
    Trophy Points:
    1
    No, it's not finite element analysis. It's DOE and mathematical modelling.

    Picture 1: DOE
    Picture 2: 2-way interaction effect estimates
    Picture 3: 3-way and i don't see anything.
     

    Attached File(s): 1. Scan for viruses before using. 2. Report any 'bad' files by reporting this post. 3. Use at your own Risk.:

    • 11.jpg
      11.jpg
      File size:
      217.9 KB
      Views:
      4
    • 13.jpg
      13.jpg
      File size:
      221.2 KB
      Views:
      4
    • 12.jpg
      12.jpg
      File size:
      99.5 KB
      Views:
      4
  6. Miner

    Miner Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    264
    Likes Received:
    186
    Trophy Points:
    42
    Location:
    Greater Milwaukee USA
    The model is saturated, so without replicates or multiple center points, you cannot estimate the 3-way interaction. The 2-way interactions are not significant. If you remove them from the model, you will find that the three factors are significant. The R^2(adj) = 88% and R^2(pred) = 73%
     
    Antisa93 and Atul Khandekar like this.
  7. Antisa93

    Antisa93 New Member

    Joined:
    Dec 2, 2016
    Messages:
    4
    Likes Received:
    0
    Trophy Points:
    1
    Yes, i understand now. Thank you very much.
     

Share This Page