delay_model: Fit a delayed Exponential or Weibull model to one or two...

In incubate: Parametric Time-to-Event Analysis with Variable Incubation Phases

Fit a delayed Exponential or Weibull model to one or two given sample(s).

Description

Maximum product of spacings estimation is used by default to fit the parameters. Estimation via naive maximum likelihood (⁠method = 'MLEn⁠) is available, too, but MLEn yields biased estimates. MLEc is a corrected version of MLE due to Cheng.

Usage

delay_model(
  x = stop("Specify observations for at least one group x=!", call. = FALSE),
  y = NULL,
  distribution = c("exponential", "weibull"),
  twoPhase = FALSE,
  bind = NULL,
  ties = c("density", "equidist", "random", "error"),
  method = c("MPSE", "MLEn", "MLEw", "MLEc"),
  profiled = method == "MLEw",
  optim_args = NULL,
  verbose = 0
)

Arguments

x	numeric. observations of 1st group. Can also be a list of data from two groups.
y	numeric. observations from 2nd group
distribution	character. Which delayed distribution is assumed? Exponential or Weibull.
twoPhase	logical. Allow for two phases?
bind	character. parameter names that are bind together in 2-group situation.
ties	character. How to handle ties.
method	character. Which method to fit the model? 'MPSE' = maximum product of spacings estimation or 'MLEn' = naive maximum likelihood estimation or 'MLEw' = weighted MLE' or MLEc' = corrected MLE
profiled	logical. Profile out scale from log-likelihood if possible.
optim_args	list. optimization arguments to use. Use NULL to use the data-dependent default values.
verbose	integer. level of verboseness. Default 0 is quiet.

Details

Numerical optimization is done by stats::optim.

Value

incubate_fit the delay-model fit object. Or NULL if optimization failed (e.g. too few observations).

incubate documentation built on Sept. 11, 2024, 6:50 p.m.