confInt: Confidence intervals for pairwise EC estimates

In gremes: Estimation of Tail Dependence in Graphical Models

Confidence intervals for pairwise EC estimates

Description

Provides confidence intervals for the estimates using the ECE. It is suitable when only pairs are used - pairwise extremal coefficients.

Usage

confInt(obj, ...)

## S3 method for class 'EKS'
confInt(obj, evpo, k, level = 0.05, ...)

Arguments

obj	is an object of class EKS which contains the estimates of the edge weights
...	additional arguments
evpo	is the matrix of evaluation points used in the EC estimator. It should contain only pairs of ones.
k	is the number of upper order statistics. It should be the same as in the estimates of the edge weights.
level	is the level of confidence. default is 0.05.

Value

A matrix with two columns for the lower and upper bounds for each of the coefficients.

Examples

seg<- make_tree(3, mode = "undirected")
seg<- set.vertex.attribute(seg, "name", V(seg), letters[1:3])
X<- matrix(rnorm(5*3), 5, 3) # create the dataset just to create the Tree object
colnames(X)<- letters[1:3]
tobj<- Tree(seg, X)
eks<- EKS(seg)
# assign any parameters, in practice these should be the estimates from the ECE
eks<- setParams(eks, c(0.1, 0.2, 0.3))
tup<- Tuples()
x<- rep(1, vcount(seg))
names(x)<- getNodesWithData(tobj)
coord<- evalPoints(tup, tobj, x)
# suppose that the estimates in deParams are obtained for k=100
confInt(eks, coord, 100)

gremes documentation built on Feb. 16, 2023, 8:06 p.m.