# complete.log.likelihood_q0: Complete log-likelihood function for xUx models and q=0 In fabMix: Overfitting Bayesian Mixtures of Factor Analyzers with Parsimonious Covariance and Unknown Number of Components

## Description

Complete log-likelihood function for models with different error variance per component (xUx) and diagonal covariance structure per component (q=0.

## Usage

 1 complete.log.likelihood_q0(x_data, w, mu, SigmaINV, z) 

## Arguments

 x_data n\times p matrix containing the data w a vector of length K containing the mixture weights mu K\times p matrix containing the marginal means per component SigmaINV K\times p\times p precision matrix (inverse covariance) per component z A vector of length n containing the allocations of the n datapoints to the K mixture components

## Value

complete log-likelihood value

## Author(s)

Panagiotis Papastamoulis

## Examples

  1 2 3 4 5 6 7 8 9 10 11 12 13 14 15  library('fabMix') data(waveDataset1500) x_data <- scale(waveDataset1500[ 1:20, -1]) # data z <- waveDataset1500[ 1:20, 1] # class p <- dim(x_data)[2] q <- 2 K <- length(table(z)) # 3 classes # give some arbitrary values to the parameters: set.seed(1) w <- rep(1/K, K) mu <- array( -0.1 + 0.2*runif(K * p), dim = c(K,p) ) SigmaINV <- array( 1, dim = c(K,p,p)) # compute the complete.log.likelihood ( -inf ) complete.log.likelihood_q0(x_data = x_data, w = w, mu = mu, SigmaINV = SigmaINV, z = z) 

fabMix documentation built on Feb. 20, 2020, 1:09 a.m.