envlpaster (version 0.1-2)

scanner: scanner

Description

A diagnostic assessing the potential benefits of envelope methodology for a particular aster model.

Usage

scanner(M, coef, u)

Arguments

M

A $\sqrt{n}$ estimate of an estimator's asymptotic covariance matrix.

coef

The MLE of the parameter of interest.

u

The dimension of the envelope space assumed.

Value

indices

The indices of the u most relevant eigenvectors of M to the construction of coef.

table

An output table. The first column is the projection of coef into the space spanned by the eigenvectors given by the indices.

G

The u most relevant eigenvectors of M to the construction of coef.

prop

the sum of the remaining eigenvalues of M divided by the sum of all of the eigenvalues of M.

Details

This function provides users with a rough diagnostic for the performance of an envelope estimator at a specific dimension. We can see how close a particular potential envelope estimator is to the MLE as well as the proportion of variation that will be discarded when using envelope estimation. This amount of variation discarded is optimistic since it does not account for variability associated with estimating the projection into the envelope space.

Examples

## Not run: library(envlpaster)
# data(simdata30nodes)
# data <- simdata30nodes.asterdata
# nnode <- length(vars)
# xnew <- as.matrix(simdata30nodes[,c(1:nnode)])
# m1 <- aster(xnew, root, pred, fam, modmat)
# avar <- m1$fisher
# beta <- m1$coef
# scanner(M = avar, coef = beta, u = 1)## End(Not run)