compositemat: Estimation of the variability and Godambe information matrix...

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

Description

Composite likelihood estimators are consistent and asymptotically normal under mild regularity conditions. The asymptotic covariance matrix is the inverse Godambe information matrix G^{-1}, which is estimated by a nonparametric bootstrap as the empirical covariance of B replicates. The sensitivity matrix H is calculated from the Hessian matrix at the maximum composite likelihood estimates. Lastly, the variability matrix J is obtained from the relation G=HJ^{-1}H

Usage

compositemat(dat, fitted, B, use.start = FALSE, ...)

Arguments

dat	data matrix
fitted	output from the call to fmvcpot.
B	number of bootstrap replicates
use.start	logical indicating whether to use MCLE as starting value
...	fixed pararameters to pass to fmvcpot if any

Value

a list with matrices godambe, sensitivity and variability

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