[R] estimation stable distribution parameters
Jim Lindsey
james.lindsey at luc.ac.be
Fri Jun 6 11:17:36 CEST 2003
>
> Hi,
> This just repetition of my mail, so please apologize me if you receive it
> twice.
>
> I am wondering whether anyone of you already implemented in R or S+ the
> methods of estimation the parameters of stable distribution (especially the
> iid case). If there is, i will be happy to get the copy of the codes.
>
> Some R-users referred me to the package stable (thanks M. Maechler and B.
> Pikounis), but i think it does not give the answer to my question. Probably
> i can use function stablereg in the package, but after read the explanation
> of the function, i am totally confused with the result reported. Simply
> applying what they shown in the example in the help for this function, for
> example check for the line after
>
> #Stable model with loc(ation)=loc.h(b0+b1*day)
>
> i got result of skewness estimate as -2.05 which is an incorrect thing
> (? i am not sure, probably my interpretation is wrong)
It is correct for the parametrisation used.
>
> It looks like they implemented the method of estimation for regression
> model based on approximate / conditional maximum likelihood (I didn't check
> their codes yet, and also until now i can not get the referred paper). I
> intended to see the accuracy of the other estimation procedures e.g. the
> method of quantile's of McCulloch or the method based on ECF. Some
> colleagues referring me to Xplore, where Weron has implement the methods
> using the languange Xplore, however it is inaccessible for public.
>
It uses exact MLE with numerical precision set by the integration
(user specified). The methods to which you refer are all obsolete.
Jim
> Please let me know if any body has any clues to my question. Otherwise, i
> have to start writing my own codes
>
> Regards
> Dedi t
>
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://www.stat.math.ethz.ch/mailman/listinfo/r-help
>
More information about the R-help
mailing list