hi!
i've tried to run this code in R using the VGAM package. I know it's
not a good fit. But i encountered a problem in calling the estimated parameter
values using fit$estimate. But the codes worked if i put the estimated values
just like in this case:
y.l <- dgpd(x, scale=0.979421685 , shape=-0.003183445 )
But not in this one:
y.g <- dgpd(x, scale=fit$estimate[1], shape=fit$estimate[2])
Can anyone help me how to call out the estimated parameters automatically
instead of writing each of the estimated values? So i can use them in future
graphing. I know this will cause a big problem if i'll keep on changing my
data. Further, it will be a time constraint problem if i'll fit
distributions to data using many VGAM distributions. Please see the R code
below for reference.
Thank you for the help!
Filame
----------------------------------------------------------------------------------------------------
library(VGAM)
z<-rgamma(10000,rate=1, shape=1)
fit <- vglm(z~1, gpd(threshold=0),trace=TRUE)
coef(fit, mat=TRUE)
Coef(fit)
x<-seq(from=0, to=40, by=1)
y.l <- dgpd(x, scale=0.979421685 , shape=-0.003183445 )
truehist(z)
lines(x, y.l, col=2)
y.g <- dgpd(x, scale=fit$estimate[1], shape=fit$estimate[2])
truehist(z)
lines(x, y.g, col=2)
ERROR:
Error in dgpd(x, scale = fit$estimate[1], shape = fit$estimate[2]) :
"scale" must be positive
In addition: Warning message:
$ operator not defined for this S4 class, returning NULL in: fit$estimate
--------------------------------------------------------------------------------------------------------
---------------------------------
[[alternative HTML version deleted]]
