Profile log likelihood for extended generalized Pareto models

Computes the profile log likelihood over a grid of values of \(\psi\) for various parameters, including return levels.

Usage

egp.pll(
  psi,
  model = c("pt-beta", "pt-gamma", "pt-power", "gj-tnorm", "gj-beta", "exptilt",
    "logist"),
  param = c("kappa", "scale", "shape", "retlev"),
  mle = NULL,
  xdat,
  thresh = NULL,
  plot = FALSE,
  method = c("Nelder", "nlminb", "BFGS"),
  p,
  ...
)

Arguments

psi: grid of values for the parameter to profile
model: string; choice of extended eneralized Pareto model.
param: string; parameter to profile
mle: a vector or matrix with maximum likelihood estimates of kappa, scale, shape. This can be a matrix if there are multiple threshold
xdat: vector of observations
thresh: vector of positive thresholds. If NULL, defaults to zero.
plot: logical; if TRUE, returns a plot of the profile log likelihood
method: string giving the optimization method for the outer optimization in the augmented Lagrangian routine; one of nlminb or BFGS
p: tail probability for return level if param="retlev".
...: additional arguments, currently ignored

Value

an object of class eprof