em_icl: Integrated Complete-data Likelihood (ICL) Criterion

In betaclust: A Family of Beta Mixture Models for Clustering Beta-Valued DNA Methylation Data

Integrated Complete-data Likelihood (ICL) Criterion

Description

Compute the ICL value for the optimal model.

Usage

em_icl(llk, C, M, N, R, model_name = "K..", z)

Arguments

llk	Log-likelihood value.
C	Number of CpG sites.
M	Number of methylation states identified in a DNA sample.
N	Number of patients.
R	Number of DNA sample types collected from each patient.
model_name	Fitted mixture model. Options are "K..", "KN." and/or "K.R" (default = "K..").
z	A matrix of posterior probabilities of cluster membership, stored as z in the object from beta_k, beta_kn and beta_kr functions.

Details

Computes the ICL for a specified model given the log-likelihood, the dimension of the data, and the model names. This criterion penalises the BIC by including an entropy term favouring well separated clusters.

Value

The ICL value for the selected model.

See Also

em_aic

em_bic

betaclust documentation built on Sept. 30, 2024, 9:30 a.m.