GRMoE-class: A Reference Class which contains parameters of a GRMoE model.

In fchamroukhi/HDME: LASSO Regularized Mixture of Experts Models

Description

GRMoE contains all the parameters of a Gaussian Regularized Mixture-of-Experts.

Fields

X

The matrix data for the input.

Y

Vector of the response variable.

d

Numeric. Number of explanatory variables (including the intercept variable).

n

Numeric. Length of the response/output vector Y.

K

Number of expert classes.

Lambda

Penalty value for the expert part.

Gamma

Penalty value for the gating network.

wk

Parameters of the gating network. Matrix of dimension (K - 1, d), with d the number of explanatory variables (including the intercept).

betak

Regressions coefficients for each expert. Matrix of dimension (d, K).

sigma

Numeric. The standard deviation.

loglik

Numeric. Observed-data log-likelihood of the GRMoE model.

storedloglik

Numeric vector. Stored values of the log-likelihood at each EM iteration.

BIC

Numeric. Value of BIC (Bayesian Information Criterion).

zerocoeff

Matrix. Proportion of zero coefficients obtained during each iteration of the EM. First column gives the number of zero coefficients for wk and the second column for betak.

Cluster

Numeric. Clustering label for each observation.

Methods

plot(what = c("loglik", "zerocoefficients"))

Plot method.

what

The type of graph requested:

• "loglik" = Value of the log-likelihood for each iteration.

• "zerocoefficients" = Proportion of zero coefficients for each iteration.

By default, all the above graphs are produced.

fchamroukhi/HDME documentation built on Nov. 4, 2019, 12:37 p.m.