GLENV: Generalized Linear Envelopes for Sufficient Dimension...

In abnormally-distributed/cvreg: Cross Validation and Robust Estimation Utilities

Description

This is an adaptation with several improvements on functions in the Renvlp package. Incorporated here are methods for fitting predictor envelopes for binomial and poisson regression with Firth's biased reduced estimation process, which removes the second-order bias in the maximum likelihood estimate. This also improves the resilience of the estimator when there is class separation, in which case the MLE is degenerate.

Usage

GLENV(
  formula,
  data,
  rank = "all",
  family = c("binomial", "poisson"),
  se = TRUE,
  init = NULL
)

Arguments

formula	model formula
data	data frame
rank	number of dimensions
se	should standard errors for the regression basis be calculated? defaults to TRUE.
init	an optional initial covariate dimension reduction subspace. defaults to NULL.

Value

an sdr object

abnormally-distributed/cvreg documentation built on May 3, 2020, 3:45 p.m.