deGEM: Joint estimation of differential networks.

In drjingma/deGEM: Joint estimation of multiple differential networks

Description

Joint estimation of differential networks.

Usage

deGEM(z, K, lambda, maxIter = 50, tol = 0.001)

Arguments

z	An N by p data matrix.
K	Number of components in the mixture model.
lambda	A vector of tuning parameters.
maxIter	The maximum number of iterations in EM
tol	The tolerance level for EM convergence

Details

There are several issues with the current form. First, this function may produce poor estimates if the penalty parameters are too large. Unfortunately, when the penalty parameter is small, the estimates of the differential network may not be sparse. In addition, the class labels may have swapped during the initialization, so it is important to reevaluate the basis class.

Value

A list with the elements

pie	The mixing proportions.
mu	The mean of each class with dimension p x K.
D	The differential networks with dimension p x p x K.
Omega	The precision matrix of each class with dimension p x p x K.
AIC	The AIC for model selection.
BIC	The BIC for model selection.

drjingma/deGEM documentation built on May 25, 2019, 4:24 p.m.