R help - Jan 2008 - Problem with estimation of distribution parameters

Hi, 

 

I tried to calculate the MLE estimation of a parameter of a gamma
distribution of a data set, which is called "amount".. I did it with
the
command: fitdistr(amount,"gamma") 

and got the error message:

Error in optim(x = c(363.36, 551.15, 603.91, 170.99, 225.21, 218.01,
1816.82,  : 

        non-finite finite-difference value [2]

Warning message:

NaNs wurden erzeugt in: dgamma(x, shape, scale, log)

 

What does it mean? What can I do?

 

Kind regards, 

Oliver

 


	[[alternative HTML version deleted]]

Oliver Herbst <oliverherbst <at> gmx.de> writes:
> 
> Hi, 
> 
> I tried to calculate the MLE estimation of a parameter of a gamma
> distribution of a data set, which is called "amount".. I did it
with the
> command: fitdistr(amount,"gamma") 
> 
> and got the error message:
> 
> Error in optim(x = c(363.36, 551.15, 603.91, 170.99, 225.21, 218.01,
> 1816.82,  : 
> 
>         non-finite finite-difference value [2]
> 
> Warning message:
> 
> NaNs wurden erzeugt in: dgamma(x, shape, scale, log)
> 
> What does it mean? What can I do?
> 
    Send (or post somewhere if it is too large) the full data set (amount).
(See the posting guide: we like REPRODUCIBLE examples ...)

  cheers
     Ben Bolker