I ran the response surface modeling using central composite design. And I got this output. Can I take this model? The model is having lack of fit p value more than 0.05. But other parameters are not good. Such as the p values in x1 , x2 and x1:x2 are non-significant. The model gives the stationary points at pH=6.7 and Temp= 21.8. Are these stationary points are reliable given the structure of model. The output is attached herewith. I used the rsm package of R
Because you are building A predictive model p-values are not of much help. I would look more at things that tell you if your residuals are small. Of course you could also use adaptive lasso variable selection which, because of it's oracle property yields an optimal model given your candidate predictors. See:
https://www4.stat.ncsu.edu/~boos/var.select/lasso.adaptive.html . I expect we will see more of adaptive lasso in RSM once it's properties are are more widely known.
Best, David Booth
Since all of the effects in your model are insignificant so the stationary points are not reliable.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics