meteorits (version 0.1.1)

StatStMoE-class: A Reference Class which contains statistics of a StMoE model.

Description

StatStMoE contains all the statistics associated to a StMoE model. It mainly includes the E-Step of the ECM algorithm calculating the posterior distribution of the hidden variables, as well as the calculation of the log-likelhood.

Fields

piik

Matrix of size \((n, K)\) representing the probabilities \(\pi_{k}(x_{i}; \boldsymbol{\Psi}) = P(z_{i} = k | \boldsymbol{x}; \Psi)\) of the latent variable \(z_{i}, i = 1,\dots,n\).

z_ik

Hard segmentation logical matrix of dimension \((n, K)\) obtained by the Maximum a posteriori (MAP) rule: \(z\_ik = 1 \ \textrm{if} \ z\_ik = \textrm{arg} \ \textrm{max}_{s} \ \tau_{is};\ 0 \ \textrm{otherwise}\), \(k = 1,\dots,K\).

klas

Column matrix of the labels issued from z_ik. Its elements are \(klas(i) = k\), \(k = 1,\dots,K\).

tik

Matrix of size \((n, K)\) giving the posterior probability \(\tau_{ik}\) that the observation \(y_{i}\) originates from the \(k\)-th expert.

Ey_k

Matrix of dimension (n, K) giving the estimated means of the experts.

Ey

Column matrix of dimension n giving the estimated mean of the StMoE.

Var_yk

Column matrix of dimension K giving the estimated means of the experts.

Vary

Column matrix of dimension n giving the estimated variance of the response.

loglik

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

com_loglik

Numeric. Complete-data log-likelihood of the StMoE model.

stored_loglik

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

BIC

Numeric. Value of BIC (Bayesian Information Criterion).

ICL

Numeric. Value of ICL (Integrated Completed Likelihood).

AIC

Numeric. Value of AIC (Akaike Information Criterion).

log_piik_fik

Matrix of size \((n, K)\) giving the values of the logarithm of the joint probability \(P(y_{i}, \ z_{i} = k | \boldsymbol{x}, \boldsymbol{\Psi})\), \(i = 1,\dots,n\).

log_sum_piik_fik

Column matrix of size m giving the values of \(\textrm{log} \sum_{k = 1}^{K} P(y_{i}, \ z_{i} = k | \boldsymbol{x}, \boldsymbol{\Psi})\), \(i = 1,\dots,n\).

dik

It represents the value of \(d_{ik}\).

wik

Conditional expectations \(w_{ik}\).

E1ik

Conditional expectations \(e_{1,ik}\).

E2ik

Conditional expectations \(e_{2,ik}\).

E3ik

Conditional expectations \(e_{3,ik}\).

stme_pdf

Skew-t mixture of experts density.

Methods

computeLikelihood(reg_irls)

Method to compute the log-likelihood. reg_irls is the value of the regularization part in the IRLS algorithm.

computeStats(paramStMoE)

Method used in the ECM algorithm to compute statistics based on parameters provided by the object paramStMoE of class ParamStMoE.

EStep(paramStMoE, calcTau = FALSE, calcE1 = FALSE, calcE2 = FALSE, calcE3 = FALSE)

Method used in the ECM algorithm to update statistics based on parameters provided by the object paramStMoE of class ParamStMoE (prior and posterior probabilities).

MAP()

MAP calculates values of the fields z_ik and klas by applying the Maximum A Posteriori Bayes allocation rule.

\(z_{ik} = 1 \ \textrm{if} \ k = \textrm{arg} \ \textrm{max}_{s} \ \tau_{is};\ 0 \ \textrm{otherwise}\)

See Also

ParamStMoE