estimate: Provides estimation of parameters

In gremes: Estimation of Tail Dependence in Graphical Models

Provides estimation of parameters

Description

Estimates the edge weights or edge parameters of a model parameterized cliquewise by Huesler-Reiss distributions. For description of the model, please consult the Vignette "Introduction".

Usage

estimate(obj, ...)

## S3 method for class 'HRMtree'
estimate(obj, Data, subsets, k_ratio, xx = NULL, ...)

## S3 method for class 'HRMBG'
estimate(obj, Data, subsets, k_ratio, xx = NULL, ...)

## S3 method for class 'EKS'
estimate(obj, Data, coord, k_ratio, ...)

## S3 method for class 'MMEave'
estimate(obj, Data, k_ratio, ...)

## S3 method for class 'MLEave'
estimate(obj, Data, k_ratio, ...)

Arguments

obj	object of on of the following classes MME, MLE1, MLE2, EKS, EKS_part, EngHitz, MME_ave, MLE_ave, HRMBG. The classes MME, MLE1, MLE2, EKS, EKS_part, EngHitz are subclasses of class HRMtree, hence if no other method is available, the method for HRMtree applies.
...	additional arguments
Data	the dataset, it should have named columns.
subsets	object of class RootDepSet containing the roots and the respective subsets. When obj is of class EngHitz the argument subsets should be a list of subsets, not a RootDepSet object. When obj is of class EngHitz the union of all subsets should cover the whole vertex set and also every pair of subsets can have at most one node in common. See also Vignettes "Code - Note 4" and "Estimation - Note 4".
k_ratio	a scalar the ratio of the number of upper order statistics to the number of observations
xx	A vector of length the number of observed variables and named with the names of the nodes with observed variables. Default is NULL. For estimate.EKS_part it should be filled in. For the other methods the value at default should be kept.
coord	A matrix of evaluation points with number of columns equal to the number of nodes in the tree and column names according to the names of the nodes. The matrix of evaluation points should not contain a positive coordinate for a variable which is unobserved. For creation of coordinates based on tuples, triples, and adjacent nodes see classes Coordinates, Tuples, Triples, Adjacent as well as method evalPoints

Details

For an object of class EKS_part for a fixed subset all coordinates based on both tuples and triples are taken.

Value

An object of the same class as the argument obj where the slot $depParams are the estimates of edge weights. For objects MME it returns the squared values of the parameters.

Note

For detailed examples please consult Vignettes "Code - Note" 1-6.

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