I am doing a RSM (CCD) design to optimize my works based on the two response variables (Y1 and Y2).The Y1 is linear and the Y2 is quadratic. In this case, how do I optimize my works?
I am confused, because I have conducted the experiments at corner points, axial points and center points. When I try to optimize them by using RSM (CCD) the Y1 is not fit to the model due to the lack of fit is significant, Prob>F is lower than 0.05 (it should be linear as when conducting linear regression the R-square is higher and the lack of fit is insignificant) however the Y2 is fit well to the model that developed in RSM due to high R-square and the lack of fit is insignificant (Prob>F is higher 0.05).
What you are describing is a multiobjective optimization. JMP is smart enough to know that you have 2 different models. I don't know exactly how to do it in JMP but, there will be an option to optimize both models simultaneously.
As far as Y1 goes, have you tried using a transformation on your data? Are you using the proper method of analysis for Y1? What is your response?
I personally advice you to go through the study of the effect parameters in terms linear and exclude non significant parameters and calculate linearly thereafter taking into account the quadratic effect and keeping the cross-terms,
Dear Engr Low Jiun Hor
Regardless the type of model (linear and / or quadratic even power model) you can solve your multi-objective optimization problem using RSM under MINITAB.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Applied Statistics