bc.twocomp: bc.twocomp

In bimixt: Estimates Mixture Models for Case-Control Data

Description

Implementation of two component models. In the two component unconstrained model, the components of the control and case mixtures are the same; however the mixture probabilities may differ for cases and controls. In the two component constrained model, all controls are distributed according to one of the two components while cases follow a mixture distribution of the two components.

Usage

bc.twocomp(x.cases, x.controls, constrained = T, lambda.bounds = c(-5, 5),

control.comp = 1, start.vals=NULL)

Arguments

x.cases	a numeric vector of case values
x.controls	a numeric vector of control values
constrained	Boolean indicating whether the two component constrained model should be used (default T) or the two component unconstrained model should be used (F)
lambda.bounds	numeric vector of bounds: c(upper bound, lower bound). Specifies the range for optim to search for the optimization of lambda. Default: c(-5,5).
control.comp	indicator of which component contains the controls (1 or 2)
start.vals	starting values for the EM algorithm. If NA, the starting values are estimated from the data.

Value

lambda	Box-Cox transformation parameter
type	model type ( "2cc" or "2cu")
mu.cases	means of the Box-Cox transformed case components
sig.cases	standard deviations of the Box-Cox transformed case components
pi.cases	proportion of cases in each case component
mu.controls	means of the Box-Cox transformed control components
sig.controls	standard deviations of the Box-Cox transformed control components
pi.controls	proportion of controls in each control component (always equal to 1 for 2cc since all controls are forced into one component)
max.loglike	the maximum log likelihood value for the model
mu.cases.unt	an estimate of the untransformed means of the case components. Based on Monte Carlo simulations. Values will differ by computer seed.
sig.cases.unt	an estimate of the untransformed standard deviations of the case components. Based on Monte Carlo simulations. Values will differ by computer seed.
mu.controls.unt	an estimate of the untransformed means of the control components. Based on Monte Carlo simulations. Values will differ by computer seed.
sig.controls.unt	an estimate of the untransformed standard deviations of the control components. Based on Monte Carlo simulations. Values will differ by computer seed.
case	original case values
control	original control values
time	running time for the model fit

Author(s)

Michelle Winerip, Garrick Wallstrom, Joshua LaBaer

See Also

bc.binorm bc.fourcomp em.twocomp.m1 em.twocomp.m2 em.twocomp.m3

bimixt documentation built on May 2, 2019, 3:31 p.m.