Message-ID: <D736F454355006429DA41754A53FFDBB43CD1107@exmbx06-cdc.nexus.csiro.au>
Date: 2015-05-27T09:06:34Z
From: Onoriode.Coast at csiro.au
Subject: Estimating Four Parameters with nls in R
Hello all!
I am trying to estimate four parameters of a model (mu, sigma, theta and lambda) for a subset of data (snapshot given below). I have tried to do this using the nls package in R but I kept getting error messages (Error in numericDeriv(form[[3L]], names(ind), env) : --AND-- Missing value or an infinity produced when evaluating the model). Please have a look at my code and tell, if you can, how I might get it to work. The data has five levels of "WP" (0, -0.4, -0.8, -1.2 and -1.6).
Mod1<-nls(1-germ~exp(-1*((WP-(theta/time)-mu)/sigma)^lambda),start=c(mu=-3,theta=3,sigma=3,lambda=-1),data=ht)
WP
time
germ
0
1.333333
0
0
1.416667
0
0
1.5
0.04
0
1.583333
0.04
0
2.083333
0.08
0
2.166667
0.16
0
2.25
0.24
0
2.333333
0.36
0
2.416667
0.6
0
2.5
0.64
0
2.583333
0.64
0
2.666667
0.72
0
2.916667
1
.
.
.
-1.6
2.916667
0
-1.6
3.166667
0
-1.6
3.666667
0
-1.6
4.166667
0
-1.6
4.666667
0
-1.6
5.166667
0
-1.6
5.666667
0
-1.6
7.666667
0
-1.6
9.666667
0
-1.6
12.66667
0
-1.6
19.66667
0
Dr Onoriode Coast
Postdoctoral Fellow
Agriculture Flagship
CSIRO
E onoriode.coast at csiro.au
[[alternative HTML version deleted]]