EMMalpha: ECM: M-Step for logit regression coefficients 'alpha'.

In sparktseung/LRMoE: Logit-Weighted Reduced Mixture-of-Experts

ECM: M-Step for logit regression coefficients alpha.

Description

ECM: M-Step for logit regression coefficients alpha.

Usage

EMMalpha(X, alpha, comp.zkz.e.list, alpha.iter.max, penalty, hyper.alpha)

Arguments

X	A N*P matrix of numerical covariates.
alpha	A g*P matrix of old logit regression coefficients.
comp.zkz.e.list	An object returned by EMEzkz.
alpha.iter.max	Numeric: maximum number of iterations.
penalty	TRUE/FALSE, which indicates whether penalty is applied.
hyper.alpha	A numeric of penalty applied to alpha.

Value

alpha.new Updated logit regression coefficients.

sparktseung/LRMoE documentation built on March 21, 2022, 3:22 a.m.