[R] MASS fitdistr with plyr or data.table?

[Hadley Wickham]

On Wed, Apr 27, 2011 at 3:55 PM, Justin Haynes <jtor14 at gmail.com> wrote:
> I am trying to extract the shape and scale parameters of a wind speed
> distribution for different sites.  I can do this in a clunky way, but
> I was hoping to find a way using data.table or plyr.  However, when I
> try I am met with the following:
>
> set.seed(144)
> weib.dist<-rweibull(10000,shape=3,scale=8)
> weib.test<-data.table(cbind(1:10,weib.dist))
> names(weib.test)<-c('site','wind_speed')
>
> fitted<-weib.test[,fitdistr(wind_speed,'weibull'),by=site]
>
> Error in class(ans[[length(byval) + jj]]) = class(testj[[jj]]) :
>  invalid to set the class to matrix unless the dimension attribute is
> of length 2 (was 0)
> In addition: Warning messages:
> 1: In dweibull(x, shape, scale, log) : NaNs produced
> ...
> 10: In dweibull(x, shape, scale, log) : NaNs produced
>
> (the warning messages are normal from what I can tell)
>
> or using plyr:
>
> set.seed(144)
> weib.dist<-rweibull(10000,shape=3,scale=8)
> weib.test.too<-data.frame(cbind(1:10,weib.dist))
> names(weib.test.too)<-c('site','wind_speed')
>
> fitted<-ddply(weib.test.too,.(site),fitdistr,'weibull')

Well fitdistr doesn't return a data frame, so you need to do something
to its output...

Hadley


-- 
Assistant Professor / Dobelman Family Junior Chair
Department of Statistics / Rice University
http://had.co.nz/