[R] fitting a theoretical distribution with truncated tails

[Spencer Graves]

      Have you considered "fitdistr"?  The documentation says that the 
second argument is "Either a character string or a function returning a 
density evaluated at its first argument."  It should be easy enough to 
write something like the following: 

dweibull.trunc <-
function(x, shape, scale=1, trunc.=Inf, log=FALSE){
    ln.dens <- (dweibull(x, shape, scale, log=TRUE)
        -pweibull(trunc., shape, scale = 1, lower.tail = TRUE, log.p = 
TRUE))
    if(any(oops <- (x>trunc.)))
        ln.dens[oops] <- (-Inf)   
    if(log)ln.dens else exp(ln.dens)
}

x <- rweibull(100, 1)
range(x)
x4 <- x[x<=4]
fitdistr(x4, dweibull.trunc, start=list(shape=1, scale=1), trunc=4)

      If you want to estimate the truncation point, that will be a more 
difficult problem.  For that, I suggest you compute the max of your data 
and parameterize the truncated density with a parameter like 
"log.trunc.over.max" so "trunc." in the above example is computed as 
(max+exp(log.trunc.over.max)). 

      hope this helps.  spencer graves

Piyush Sharma wrote:

>Hi,
>  I have recently started working with R and am not really fluent in it. I
>am plotting a few graphs using the qqplot function. Is there a method for
>fitting a theoretical distribution (e.g Weibull) with truncated tails in R?
>
>Thanks for any help!
>Piyush
>
>______________________________________________
>R-help at stat.math.ethz.ch mailing list
>https://www.stat.math.ethz.ch/mailman/listinfo/r-help
>  
>