mixturetobit: Perform maximum-likelihood estimation for a mixture of tobit...
In WannabeSmith/mixturetobit: Estimate Tobit Mixtures Using the E-M Algorithm

Description Usage Arguments Value

View source: R/mixturetobit.R

This function performs maximum-likelihood estimation via the E-M algorithm to obtain estimates of regression coefficients in a mixture of tobit regression models.

mixturetobit(formula, data, K = 2, start.beta = NULL,
  start.sigma = NULL, start.lambda = NULL, id = NULL, left = -1,
  tol = 1e-05, theta.lower = NULL, theta.upper = NULL,
  method = "L-BFGS-B", beta.starting.sd = 0.5)

`formula`	a regression formula describing the relationship between the response and the covariates
`data`	the data.frame containing the responses and covariates
`K`	the number of mixtures (or latent classes)
`start.beta`	a list of length K of starting values for each mixture's beta coefficients
`start.sigma`	a vector of length K of starting values for each mixture's sigma value
`start.lambda`	a vector of length K of starting values for the mixing proportions
`id`	a string specifying the name of the column that identifies subjects
`left`	a number specifying where left-censoring occurred
`tol`	a vector of numbers specifying the tolerance(s) used to determine convergence. If length(tol) > 1, the methods used should be supplied as a vector to "method".
`theta.lower`	a numeric vector of lower bounds for the theta parameters
`theta.upper`	a numeric vector of upper bounds for the theta parameters
`method`	a vector of strings specifying the optimization routine(s) to be used by optim. If length(method) > 1, optimization will be performed in succession with the tolerances provided to "tol"
`beta.starting.sd`	a numeric for what std deviation to use for starting the EM with random betas

a list containing the following elements:

`beta`	a list containing the estimated regression coefficients
`sigma`	a vector containing the estimated values of sigma
`lambda`	a vector containing the estimated mixing proportions
`delta`	a list of length K containing the estimated class membership probabilities for each observation
`ll`	the log-likelihood function evaluated at the MLE