0

I have fitted a model and estimated parameters for;

Count = beta0*exp(beta1*time)

If an additional parameter is introduced such that the model becomes;

Count = alpha + (beta)*exp(beta1*time)

I tried using the same method as i did to find just the beta0 and beta1 term using the command;

fit<-nls(count~beta*exp(beta1*time), start=list(alpha=0, beta0=500, bet1=0), data=radiation)

but get the following error... 

Error in nlsModel(formula, mf, start, wts) : 
  singular gradient matrix at initial parameter estimates

Any help would be greatly appreciated!! Thanks in advance

Tetsujin no Oni
  • 7,300
  • 2
  • 29
  • 46
user3647887
  • 1

1 Answers1

0

With your starting value of beta1=0, your equation alpha+beta0*exp(beta1*time) reduces to alpha+beta0, which means that the derivatives of the function with respect to alpha and beta0 are confounded. Try changing your starting value for beta1 to something slightly different from 0 ...

Ben Bolker
  • 211,554
  • 25
  • 370
  • 453