Gompertz MLE function — gompertz

Fits a Gompertz distribution with proportional hazards to doubly-truncated mortality data using maximum likelihood estimation.

Usage

gompertz_mle(
  formula,
  left_trunc = 1975,
  right_trunc = 2005,
  data,
  byear = byear,
  dyear = dyear,
  lower_age_bound = NULL,
  upper_age_bound = NULL,
  weights = NULL,
  start = NULL,
  death_age_data_type = "auto",
  maxiter = 10000
)

Arguments

formula: the estimation formula
left_trunc: left truncation year
right_trunc: right truncation year
data: a data frame containing variables in the model
byear: vector of birth years
dyear: vector of death years
lower_age_bound: lowest age at death to include (optional)
upper_age_bound: highest age at death to include (optional)
weights: an optional vector of individual weights
start: an optional vector of starting values for the optimizer. must be a numeric vector that exactly matches the output of get.par.start(formula, data) in length and element names.
death_age_data_type: option for handling of continuous and discrete death age variable (not yet implemented)
maxiter: maximum number of iterations for optimizer

Value

Returns a named list consisting of the following components (See stats::optim() for additional details):

starting_values

list of starting values of parameters

optim_fit

A list consisting of:

par: best estimation of parameter values
value: log likelihood
counts: number of calls to function and gradient
convergence: returns 0 if the model converged, for other values see stats::optim()
message: any other information returned by optimizer
hessian: Hessian matrix

results

A table of estimates and upper/lower bounds of the 95 percent confidence interval for the estimates. Confidence interval computed as 1.96*standard_error.

Examples

if (FALSE) {
#model hazards as function of birthplace using bunmd_demo file
gompertz_mle(formula = death_age ~ bpl_string, left_trunc = 1988, right_trunc = 2005,
data = bunmd_demo)
}