[R] optimal factorial designs
andy_liaw at merck.com
Mon Mar 6 02:42:19 CET 2006
From: Bart Joosen
> Hi All,
> recently I used Design Expert for some Design Of Experiment
> work. I was happy with the interface to select which effects
> I want to see in my experiment, and which not. For example: I
> can select of course my main effects, but also if I want to
> see interaction A:B, B:C, A:B:C,but not A:C.
Just out of curiosity, how do you make sense out of a model with A:B:C
interaction but not A:C?
> This was very interessting as you can end up with fewer runs,
> especially in cases of 10 factors with 2-3 levels.
> Is there a way to do this with R?
> I found the AlgDesign package, and I tried optFederov and
> optMonteCarlo, as you can set a formula (like:
> A+B+C+A:B+B:C+A:B:C). But while using this, I end up with to
> much runs, so I have some residuals. And sometimes I cant get
> the right interactions from my design.
> So: are there better ways to achieve the right design?
> Kind regards
> [[alternative HTML version deleted]]
> R-help at stat.math.ethz.ch mailing list
> PLEASE do read the posting guide!
More information about the R-help