On Jan 2, 2015, at 10:33 AM, Paul Hudson wrote:
> Hello all,
>
> I want to fit a tweedie distribution to the data I have.
>
> The R packages I have been able to find assume that I want to use it as
> part as of a generalized linear model.
>
> This is not the case, I want to directly fit the distribution to the data.
>
> Is there a package that allows this?
install.packages("sos")
findFn("tweedie")
#Scrolling down the resulting page, I do see a 'tweedie' package.
require(tweedie) # has the rtweedie function
y <- rtweedie(100, xi=1.5, mu=20, phi=3)
help(pack=tweedie) # Bottom of page says tweedie.profile will estimate power
res <- tweedie.profile(y~1)
#1.2 1.3 1.4 1.5 1.6 1.7 1.8
#.......Done.
str(res)
#-------------
List of 12
$ y : num [1:50] -410 -408 -406 -405 -403 ...
$ x : num [1:50] 1.2 1.21 1.22 1.24 1.25 ...
$ ht : num -396
$ L : num [1:7(1d)] -410 -399 -395 -394 -395 ...
$ xi : num [1:7] 1.2 1.3 1.4 1.5 1.6 1.7 1.8
$ xi.max : num 1.49
$ L.max : num -394
$ phi : num [1:7(1d)] 4.85 4.16 3.43 2.8 2.3 ...
$ phi.max : num 2.83
$ ci : num [1:2(1d)] 1.37 1.62
$ method : chr "inversion"
$ phi.method: chr "mle"
Seems to estimate both the power and dispersion parameters reasonably well.
>
> [[alternative HTML version deleted]]
This is a plain-text mailing list.
--
David Winsemius
Alameda, CA, USA
______________________________________________
R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
