control-parameters: Set control parameters for various purposes
In michaelfop/LCAvarsel: Variable Selection for Latent Class Analysis

Description Usage Arguments Details Value See Also Examples

Set control parameters for the EM algorithm for latent class model estimation, multinomial logistic regression estimation in the regression step, and genetic algorithm for variable selection procedure.

controlLCA(maxiter = 1e05, tol = 1e-04, nrep = 5)

controlReg(maxiter = 5000, tol = 1e-05)

controlGA(popSize = 20, maxiter = 100, run = maxiter/2,
          pcrossover = 0.8, pmutation = 0.2,
          elitism = base::max(1, round(popSize*0.05)))

`maxiter`	Maximum number of iterations in the EM algorithm, the multinomial logistic regression and the genetic algorithm.
`tol`	Tolerance value for judging when convergence has been reached. Used in the EM algorithm and the multinomial logistic regression.
`nrep`	Number of times to estimate the latent class analysis model, using different starting values for the matrix `z` of a posteriori probabilities. Each time, a different random initialization is used in order to search for the global maximum of the log-likelihood.
`popSize`	Population size. This number corresponds to the number of different models to be considered at each iteration of the genetic algorithm.
`run`	Number of consecutive generations without any improvement in the best fitness value of the variable selection procedure before the genetic algorithm is stopped.
`pcrossover`	Probability of crossover between pairs of models.
`pmutation`	Probability of mutation in a parent model.
`elitism`	Number of best fitness models to survive at each iteration of the genetic algorithm in the variable selection procedure.

Function controlLCA is used to set control parameters of the EM algorithm employed to estimate the latent class analysis model.

Function controlReg controls tolerance and maximum number of iterations in the estimation of the multinomial logistic regression. This regression is used to model the conditional distribution of a proposed variable given the current set of clustering variables in the variable selection procedure.

Function controlGA sets parameters of the genetic algorithm used for variable selection.

A list of parameters values.

poLCA, ga

1
2
3

data(carcinoma, package = "poLCA")
# increase number of replicates and decrease tolerance value
fit <- fitLCA(carcinoma, ctrlLCA = controlLCA(nrep = 10, tol = 1e-07))