In case of Response Surface Methodology with Central Composite Design design what value of alpha is preferred? What is the effect of high value of alpha on DOE?In MINITAB17 how to change alpha value.
The preferred value for alpha for a CCD is k^(1/2) where k is the number of variables
So k = 2 ; alpha = sqrt (2) ; k = 3 ; alpha = sqrt (3) ; k = 4 ; alpha = 2 ...
In fact, it is so because these values put the axial (or star) points on the same hypersphere (circle in 2D, sphere in 3D, ..) as the factorial points
hi rohit,
normal alpha is 0.01, for lower it becomes -0.01, center, 0 and higher +0.01. alpha is related to the confident level example some paper used the value of alpha 0.05 which means the confident level is 95%. regards
The preferred value for alpha for a CCD is k^(1/2) where k is the number of variables
So k = 2 ; alpha = sqrt (2) ; k = 3 ; alpha = sqrt (3) ; k = 4 ; alpha = 2 ...
In fact, it is so because these values put the axial (or star) points on the same hypersphere (circle in 2D, sphere in 3D, ..) as the factorial points
I think alpha is 4th root of k and k being the number of variables.. i-e alpha = (k)^1/4. This is mentioned in the tutorials of DOE. Please someone correct me If i am wrong.
The correct formula for "Alpha" for design of experiment can be calculated as follow:
Number of factors Factorial proportion Value of alpha (related to +/-1)
K (2)^k (2^k)^1/4
2 (2)^2 (2^2)^1/4= 1.414
3 (2)^3 (2^3)^1/4=1.682 and so on..
Based on the values of the alpha the CCD (CENTRAL COMPOSITE DESIGN ) are of various types namely rotatable, spherical , face centered and practical.
for sphreical design alpha = sqrt(k) , where k is the no of factors
for k>5 , we use alpha practical alpha = (k)^(1/4).
This puts the star points on the same hypersphere In space as that of factorial points. i.e. cirlcle in 2d , sphere in 3d
- Badges
- Science method
- Similar topics
- Methodology