LogLikelihoodNotExact: Calculate the log likelihood

In UofTActuarial/LRMoE: Logit-Weighted Reduced Mixture-of-Experts

Description

Calculate the log likelihood based on the observations, gate matrix, model matrix and exposure matrix. Keep in mind our observation matrix now have a different format

Usage

LogLikelihoodNotExact(Y, gate, model, exposure)

Arguments

Y	(array) The observation matrix with Ix4J dimentions, where each entry is a tuple of this format (tl, yl, tu, yu). Keep in mind it means there will be three dimention for this matrix, and the last dim need to be 4.
gate	(matrix) The numerical gate matrix with IxD dimentions
model	(ExpertMatrix) The JxD ExpertMatrix S6 Class which contained ExpertFunctions
exposure	(matrix) The Ix1 numerical vector which specify the exposure for each observation.

UofTActuarial/LRMoE documentation built on Feb. 18, 2022, 1:36 a.m.