henvMU

A matrix M for the non-Grassmann manifold optimization problem in 
 Cook et al. (2016)

A matrix U for the non-Grassmann manifold optimization problem in 
 Cook et al. (2016)

A given dimension of the heteroscedastic envelope space. It should 
 be an interger between \(0\) and \(r\).

A \(L\) by \(1\) vector of the number of observations in 
 each group.

The user-specified value of Gamma for the envelope subspace.

initial

Estimate the heteroscedastic envelope subspace with specified dimension.

Provides a general routine, envMU, which allows estimation of the M envelope of span(U) given root n consistent estimators of M and U. The routine envMU does not presume a model. This package implements response envelopes, partial response envelopes, envelopes in the predictor space, heteroscedastic envelopes, simultaneous envelopes, scaled response envelopes, scaled envelopes in the predictor space, groupwise envelopes, weighted envelopes, envelopes in logistic regression and envelopes in Poisson regression. For each of these model-based routines the package provides inference tools including bootstrap, cross validation, estimation and prediction, hypothesis testing on coefficients are included except for weighted envelopes. Tools for selection of dimension include AIC, BIC and likelihood ratio testing. Background is available at Cook, R. D., Forzani, L. and Su, Z. (2016) <doi:10.1016/j.jmva.2016.05.006>. Optimization is based on a clockwise coordinate descent algorithm.

henvMU: Estimate the heteroscedastic envelope subspace

Description

Usage

Arguments

Value

References