extRemes (version 2.2)

distill.fevd: Distill Parameter Information

Description

Distill parameter information (and possibly other pertinent inforamtion) from fevd objects.

Usage

# S3 method for fevd
distill(x, ...)
# S3 method for fevd.bayesian
distill(x, cov = TRUE, FUN = "mean", burn.in = 499, ...)
# S3 method for fevd.lmoments
distill(x, ...)
# S3 method for fevd.mle
distill(x, cov = TRUE, ...)

Value

numeric vector giving the parameter values, and if estimation method is MLE/GMLE, then the negative log-likelihood. If the estimation method is MLE/GMLE or Bayesian, then the parameter covariance values (collapsed with c) are concatenated to the end as well.

Arguments

x

list object returned by fevd.

...

Not used.

cov

logical; should the parameter covariance be returned with the parameters (if TRUE, they are returned as a vector concatenated to the end of the returned value).

FUN

character string naming a function to use to estimate the parameters from the MCMC sample. The function is applied to each column of the results component of the returned fevd object.

burn.in

The first burn.in values are thrown out before calculating anything from the MCMC sample.

Author

Eric Gilleland

Details

Obtaining just the basic information from the fits:

distill: The distill method function works on fevd output to obtain only pertinent information and output it in a very user-friendly format (i.e., a single vector). Mostly, this simply means returning the parameter estimates, but for some methods, more information (e.g., the optimized negative log-likelihood value and parameter covariances) can also be returned. In the case of the parameter covariances (returned if cov = TRUE), if np is the number of parameters in the model, the covariance matrix can be obtained by peeling off the last np^2 values of the vector, call it v, and using v <- matrix(v, np, np).

As with ci, only distill need be called by the user. The appropriate choice of the other functions is automatically determined from the fevd fitted object.

See Also

fevd, ci.fevd, distillery::distill

Examples

data(Fort)

fit <- fevd(Prec, Fort, threshold=0.395, type="PP", units="inches", verbose=TRUE)
fit

distill(fit)