Skip to contents

Fit excess lifetime models by maximum likelihood

Source: R/parametric.R
fit_elife.Rd

This function is a wrapper around constrained optimization routines for different models with non-informative censoring and truncation patterns.

Usage

fit_elife(
  time,
  time2 = NULL,
  event = NULL,
  type = c("right", "left", "interval", "interval2"),
  ltrunc = NULL,
  rtrunc = NULL,
  thresh = 0,
  status = NULL,
  family = c("exp", "gp", "weibull", "gomp", "gompmake", "extgp", "gppiece",
    "extweibull", "perks", "perksmake", "beard", "beardmake"),
  weights = NULL,
  export = FALSE,
  start = NULL,
  restart = FALSE,
  arguments = NULL,
  check = FALSE,
  ...
)

Arguments

time

excess time of the event of follow-up time, depending on the value of event

time2

ending excess time of the interval for interval censored data only.

event

status indicator, normally 0=alive, 1=dead. Other choices are TRUE/FALSE (TRUE for death). For interval censored data, the status indicator is 0=right censored, 1=event at time, 2=left censored, 3=interval censored. Although unusual, the event indicator can be omitted, in which case all subjects are assumed to have experienced an event.

type

character string specifying the type of censoring. Possible values are "right", "left", "interval", "interval2".

ltrunc

lower truncation limit, default to NULL

rtrunc

upper truncation limit, default to NULL

thresh

vector of thresholds

status

integer vector giving status of an observation. If NULL (default), this argument is computed internally based on type.

family

string; choice of parametric family

weights

weights for observations

export

logical; should data be included in the returned object to produce diagnostic plots? Default to FALSE.

start

vector of starting values for the optimization routine. If NULL, the algorithm attempts to find default values and returns a warning with false convergence diagnostic if it cannot.

restart

logical; should multiple starting values be attempted? Default to FALSE.

arguments

a named list specifying default arguments of the function that are common to all elife calls

check

logical; if TRUE, fit all submodels to ensure that simpler models fit worst or as well

...

additional parameters, currently ignored

Value

an object of class elife_par

Note

The extended generalized Pareto model is constrained to avoid regions where the likelihood is flat so \(\xi \in [-1, 10]\) in the optimization algorithm.

The standard errors are obtained via the observed information matrix, calculated using the hessian. In many instances, such as when the shape parameter is zero or negative, the hessian is singular and no estimates are returned.

Examples

data(ewsim, package = "longevity")
fit1 <- fit_elife(
   arguments = ewsim,
   export = TRUE,
   family = "exp")
fit2 <- fit_elife(
   arguments = ewsim,
   export = TRUE,
   family = "gp")
plot(fit1)


summary(fit1)
#> Model: exponential distribution. 
#> Sampling: interval truncated
#> Log-likelihood: -219.797 
#> 
#> Threshold: 0 
#> Number of exceedances: 179 
#> 
#> Estimates
#> scale  
#>  1.35  
#> 
#> Standard Errors
#> scale  
#> 0.108  
#> 
#> Optimization Information
#>   Convergence: TRUE 
anova(fit2, fit1)
#>     npar Deviance Df        Chisq Pr(>Chisq)
#> gp     2  439.594 NA           NA         NA
#> exp    1  439.594  1 3.419987e-05  0.9953339