[R] fitting distributions with R

[Nadja Riedwyl]

Dear all
I've got the dataset
data:2743;4678;21427;6194;10286;1505;12811;2161;6853;2625;14542;694;11491;
          14924;28640;17097;2136;5308;3477;91301;11488;3860;64114;14334
I know from other testing that it should be possible to fit the data with the 
exponentialdistribution. I tried to get parameterestimates for the 
exponentialdistribution with R, but as the values 
of the parameter are very close to 0 i get into troubles. Do you know, what i 
could do in order to get estimates?How do you choose the starting values? in 
my opinion it should be around 1/mean(data).

 
#Parameterestimation  with mle() with the log-likelihood funktion of the  
#exponentialdistribution
library(stats4) 
ll<-function(beta) 
{n<-24 
x<-data2
-n*log(beta)+beta*sum(x)} 
est<-mle(minuslog=ll, start=list(beta=0.1))
summary(est) 

#instead of a result, i get:


Error in optim(start, f, method = method, hessian = TRUE, ...) :
        non-finite finite-difference value [1]
In addition: There were 50 or more warnings (use warnings() to see the first 
50)
#with fitdistr() for the exponentialdistribution
library(MASS)
fitdistr(data2,densfun=dexp,start=list(rate=0.1),lower=6e-06,method="BFGS")

#instead of a result, i get

Error in optim(start, mylogfn, x = x, hessian = TRUE, ...) :
        non-finite finite-difference value [1]
In addition: Warning messages:
1: bounds can only be used with method L-BFGS-B in: optim(start, mylogfn, x = 
x, hessian = TRUE, ...)
2: NaNs produced in: dexp(x, 1/rate, log)


i'll be very happy for any help i can get to solve this problem
thank you!