Skip to content
Prev 79454 / 398502 Next

curve fit

y ~ 1/(a + b*x^c) looks pretty good to me.

If we assume a starting value of c=1 then we can get
the starting values of a and b from lm(1/y ~ x) and
then do an nls fit.
On 10/20/05, Xiaodong Jin <close2ceo at yahoo.com> wrote: