[R] 4PL algorithm

Greg Tarpinian sasprog474 at yahoo.com
Fri Oct 13 01:36:42 CEST 2006


WinXP, Splus7 and R2.3.1.


All,

I have been using the SSfpl and SSlogis self-starting functions
in the nlme library to fit 4PL and restricted 4PL models.  I need
to adapt these routines to fit the alternative model

   f(x) = A + (B-A)/(1 + abs(x/EC50)^C)

My Question:  How do I obtain good starting values for this alternative
model?

(The pseudo-code found on pages 517 - 520 of "Mixed Effects Models in
S and Splus" is not applicable to my problem because it deals with
f(x)=A+(B-A)/(1+exp(EC50-x)/C)) which is not the same function.)


Many thanks,

      Greg



More information about the R-help mailing list