icenReg (version 1.3.5)

optCliq: Computes the MLE for a Binary Mixture Model.

Description

Computes the MLE for a Binary Mixture Model. Used internally for ICNPMLE, but may be useful in other problems. In the abstraction, solves the problem

$$\arg\max_p \displaystyle \sum_{i = 1}^n \log \left( \sum_{j = 1} p_j C_{ij} \right)$$

where $C_{ij}$ is an indicator of whether the i-th observation could have come from the j-th source. $C$ is referred to as the clique matrix.

Usage

optCliq(cliqMat, tol = 10^-10, 
        inner_loops = 100, outer_loops = 20)

Arguments

cliqMat

n x m clique matrix. n = number of observations, m = number of components

tol

numerical tolerance

inner_loops

number of inner loops used

outer_loops

number of outer loops used

Examples

testData <- simBVCen()
#simulate bivariate interval censored data

cliqMat <- MLEcens::reduc(testData, cm = TRUE)$cm
#computes the cliqMat associated with data

cliqFit <- optCliq(cliqMat)
#optimizes the component weights for clique matrix

cliqFit