profconf: Profiled Confidence interval for the GP Distribution

In POT: Generalized Pareto Distribution and Peaks Over Threshold

Profiled Confidence interval for the GP Distribution

Description

Compute profiled confidence intervals on parameter and return level for the GP distribution. This is achieved through the profile likelihood procedure.

Usage

gpd.pfshape(object, range, xlab, ylab, conf = 0.95, nrang = 100,
vert.lines = TRUE, ...)
gpd.pfscale(object, range, xlab, ylab, conf = 0.95, nrang = 100,
vert.lines = TRUE, ...)
gpd.pfrl(object, prob, range, thresh, xlab, ylab, conf = 0.95, nrang =
100, vert.lines = TRUE, ...)

Arguments

object	R object given by function fitgpd.
prob	The probability of non exceedance.
range	Vector of dimension two. It gives the lower and upper bound on which the profile likelihood is performed.
thresh	Optional. The threshold. Only needed with non constant threshold.
xlab, ylab	Optional Strings. Allows to label the x-axis and y-axis. If missing, default value are considered.
conf	Numeric. The confidence level.
nrang	Numeric. It specifies the number of profile likelihood computed on the whole range range.
vert.lines	Logical. If TRUE (the default), vertical lines are plotted.
...	Optional parameters to be passed to the plot function.

Value

Returns a vector of the lower and upper bound for the profile confidence interval. Moreover, a graphic of the profile likelihood function is displayed.

Author(s)

Mathieu Ribatet

References

Coles, S. (2001). An Introduction to Statistical Modelling of Extreme Values. Springer Series in Statistics. London.

See Also

gpd.fiscale, gpd.fishape, gpd.firl and confint

Examples

data(ardieres)
events <- clust(ardieres, u = 4, tim.cond = 8 / 365,
clust.max = TRUE)
MLE <- fitgpd(events[, "obs"], 4, 'mle')
gpd.pfshape(MLE, c(0, 0.8))
rp2prob(10, 2)
gpd.pfrl(MLE, 0.95, c(12, 25))

POT documentation built on April 14, 2022, 3:03 a.m.