s-news
[Top] [All Lists]

weighted nls

To: s-news@lists.biostat.wustl.edu
Subject: weighted nls
From: DivineSAAM@aol.com
Date: Mon, 19 May 2003 19:35:24 -0400
Dear Colleagues,

I am trying to do weighted nls. I have searched the archives and the white 
book. I tried TBS without success. The data frame and nls invocation are below.

# cs expo fit
csdata<-data.frame(
time=c(0,1,3,9,20),
resp=c(638.697,395.69,199.00,141.58,112.16)
)

# perform nlr
cs.nls<-nls(resp~A1*exp(-a1*time)+A2*exp(-a2*time),
data=csdata,
start=list(A1=700,a1=1,A2=100,a2=0.1)
)
cs.nls

I need constant coefficient of variation =5% weighting. How do I implement this?

I will summarize and post responses.

Kind Regards,

Oscar A Linares
Molecular Medicine Unit
The University of Michigan Geriatrics Center

<Prev in Thread] Current Thread [Next in Thread>