Hi,
I'm trying to fit a not linear model with the nls function to some data.
So far this is the best fitting that i found.
download.file("http://dl.dropbox.com/u/29337496/data" ,
destfile="./data")
load("data")
plot (x06Veg,y06Pop)
nlmod=nls( y06Pop ~ B + A * log(x06Veg) , start = list(A = 1, B = 1 ) )
points(x06Veg , predict(nlmod), col = 2)
co <- coef(nlmod)
f <- function(x, A, B) {B + A * log(x)}
curve(f(x,A=co["A"],B=co["B"]) , add=TRUE)
I would like to increase the curvature of the model and check if the
residual sum-of-squares decrease.
Should i just guess and change the main function (B + A * log(x06Veg))
with another one (e.g. B + A * sqrt(x06Veg)) and re-run or there is a
better way?
Thanks in Advance
Giuseppe
--
Giuseppe Amatulli
Web: www.spatial-ecology.net
[[alternative HTML version deleted]]