extRemes (version 2.2)

profliker: Profile Likelihood Function

Description

Find the profile likelihood for a range of values for an extreme value df (EVD).

Usage

profliker(object, type = c("return.level", "parameter"), xrange = NULL,
    return.period = 100, which.par = 1, nint = 20, plot = TRUE, gr = NULL,
    method = "BFGS", lower = -Inf, upper = Inf, control = list(), ...)

Value

A numeric vector is returned invisibly.

Arguments

object

A list object of class “fevd” as returned by fevd.

type

character string stating whether the parameter of interest is a regular parameter or a return level.

xrange

numeric vector of length two giving the range of values of the parameter over which to calculate the profile likelihood.

return.period

If a return level is of interest, this number gives its associated return period.

which.par

If a parameter is of interest, this number tells for which component of the parameter vector to do the profile likelihood.

nint

The profile likelihood is calculated for a sequence of nint values covering xrange.

plot

logical; should a plot of the likelihood be made? Note that this is controlled by the verbose argument in the ci method function for MLE fevd objects when “proflik” is chosen as the method for finding confidence intervals. It is usually a good idea to plot the profile likelihood to see if the confidence intervals are really found or not.

gr, method, lower, upper, control

optional arguments to optim.

...

optional arguments to plot.

Author

Eric Gilleland

Details

See the help file for ci.fevd.mle for more details on this approach.

See Also

ci.fevd.mle, fevd

Examples

z <- revd(100, loc=20, scale=0.5, shape=-0.2)
fit <- fevd(z)
fit

profliker(fit, type="parameter", which.par=3)

profliker(fit, type="parameter",
    xrange=c(-0.35, -0.2), which.par=3)