0

I have fitted different probability distributions to my data. To compare the goodness-of-fit (GOF) and decide the best distribution, I use AIC as the criteria. While I can easily extract GOF for the "fitdist" objects e.g.

fit.gamma <- fitdist(x, "gamma")
fit.norm <- fitdist(x, "norm")

aic.gamma <- fit.gamma$aic
shape.gamma <- fit.gamma$estimate[1]

and so on ...

But for

fit.gev = fevd(x ,type="GEV") # this uses fevd (from 'extRemes')

while we can display fevd's objects using 'summary(fit.gev)', there does not seem a way to extract those individual parameters (AIC or scale etc) for their use in subsequent calculations.

Any help will be greatly appreciated. Thank you.

Santosh Aryal
  • 1

1 Answers1

0

The answer is found in the extRemes manual by Eric Gilleland on page 55
https://cran.r-project.org/web/packages/extRemes/extRemes.pdf

  1. Assign summary of fevd to an object, e.g.

look1 <- summary(fit.gev.12, silent = TRUE)

  1. Extract the objects as:

AIC <- c(look1$AIC)
BIC <- c(look1$BIC)

and the parameters

location<- look1$par[1]
scale<- look1$par[2]
shape <- look1$par[3]

Done!

Santosh Aryal
  • 1