[R] how to calculace the Observational error of an variable (nls fit)

Jonas Stein news at jonasstein.de
Fri May 8 14:35:19 CEST 2009


ich have a List of measured valuepairs (w, AmpNorm) and a formula 
f(w, a, b) = AmpNorm

i used nls to find the two coefficiants a and b:

ctfit.nls <- nls(AmpNorm ~ 1/(1 + (w * a - b /w)^2) ...)

R> ctfit.nls returns
Nonlinear regression model
  model:  AmpNorm ~  1/(1 + (w * a - b /w)^2)
   data:  parent.frame() 
        a         b 
1.447e-05 9.386e+06 
 residual sum-of-squares: 0.005164

Number of iterations to convergence: 5 
Achieved convergence tolerance: 1.271e-06 

Now i want to find out the observational error. So i could write something

a = 1.4e-05 +/- 0.1e-05

or a = 1.4e-05 with 98% confidence

Thank you for reading so far and thank you for any help.

Jonas Stein <news at jonasstein.de>

More information about the R-help mailing list