fscore: Fit Dirichlet extremal model using the gradient score

In lbelzile/demoroutines: Functions for 'Extremal attractor of Liouville copulas'

Description

Optimization routine to fit a member of the extremal Dirichlet family using the gradient score

Usage

fscore(start, dat, model, p = 1, qu = 0.9, exceed = FALSE)

Arguments

start	starting values for the parameter (alpha and rho, in this order)
dat	matrix of transformed data on unit Pareto scale lying above the threshold
model	string indicating whether the "dir" or "negdir" model should be considered
p	int specifying the degree of the l_p norm used in the weight function weightFun
qu	probability level for the marginal quantile for the sum
exceed	logical indicating whether the data provided is the exceedances. Default to FALSE.

Value

the optimized parameter values

Examples

samp <- mev::rmev(n=2500,d=3,param=c(1,2,3,0.5),model="negdir")
fscore(start=c(1,2,3,0.5), dat=samp, model="negdir")

lbelzile/demoroutines documentation built on May 20, 2019, 8:28 p.m.