fmlcdEM: Estimates a Log-Concave Mixture Density

In fmlogcondens: Fast Multivariate Log-Concave Density Estimation

Description

fmlcdEM Utilizes the EM approach to obtain a mixture of log-concave densities. Utilizes Gaussian hierarchical clustering to initilize the posterior probabilities of class affiliation (as proposed by the package LogConcDEAD by Cule et al.).

Usage

fmlcdEM(X, K = 2, posterior, verbose = 0, maxIter = 50)

Arguments

X	Matrix of data points (one sample per row)
K	Number of latent variables (default: 2)
posterior	Matrix with posterior probabilities for class affiliation; Initialized if not provided using a Gaussian hierarchical clustering.
verbose	Int determining the verboseness of the code; 0 = no output to 3. (default: 0)
maxIter	Maximal number of EM iterations. (default: 50)

Value

Parametrization of the mixture density

params	List of length K, where each entry contains the hyperplane for one density
densEst	Matrix where each row contains the marginal distribution p(x)
tau	Marginal distribution over the latent variable p(z)

fmlogcondens documentation built on May 2, 2019, 8:29 a.m.