[R] optFederov question

Andras Farkas motyocska at yahoo.com
Sun Jun 8 18:55:13 CEST 2014


Dear All,

seeking input (or assurance) that I am using the optFederov in AlgDesign apropriatelly...

I have:

require(AlgDesign)
c <-seq(1,64,by=0.1)
econ=9.16
ic=4.796
hill=1.217
x=5.618

eff <-econ-(ic*(c^hill/(x^hill+c^hill)))

cand.list <-data.frame(c=c,eff=eff,econ=econ,ic=ic,hill=hill,x=x)


my goal here is to select the values in c that are most "sensitive" to most accuratelly estimating econ, ic, hill and x. These parameters will be estimated based on the relationship eff <-econ-(ic*(c^hill/(x^hill+c^hill))) using nls. To my knowledge D optimal design is one of the ways to go, thus I have started experimenting with optFederov. My qestion is if writing the code this way is optimal (or not):


optFederov(~ I(econ-(ic*(c^hill/(x^hill+c^hill)))),data=cand.list),

all of your help is greatly appreciated,

thanks,

Andras



More information about the R-help mailing list