numDeltaMethod: Calculate numerical delta method for non-linear predictions.

In rstpm2: Smooth Survival Models, Including Generalized Survival Models

Calculate numerical delta method for non-linear predictions.

Description

Given a regression object and an independent prediction function (as a function of the coefficients), calculate the point estimate and standard errors

Usage

numDeltaMethod(object, fun, gd=NULL, conf.int=FALSE, level=0.95, ...)

Arguments

object	A regression object with methods coef and vcov.
fun	An independent prediction function with signature function(coef, ...).
gd	Specified gradients
conf.int	Logical for whether to also calculate the confidence interval
level	Numeric for the level of the confidence interval
...	Other arguments passed to fun.

Details

A more user-friendly interface is provided by predictnl.

Value

fit	Point estimates
se.fit	Standard errors
Estimate	Point estimates
SE	Standard errors
conf.low	Lower confidence interval (if conf.int=TRUE)
conf.high	Upper confidence interval (if conf.int=TRUE)

See Also

See Also predictnl.

Examples

##---- Should be DIRECTLY executable !! ----
##-- ==>  Define data, use random,
##--	or do  help(data=index)  for the standard data sets.

## The function is currently defined as
function (object, fun, ...) 
{
    coef <- coef(object)
    est <- fun(coef, ...)
    Sigma <- vcov(object)
    gd <- grad(fun, coef, ...)
    se.est <- as.vector(sqrt(colSums(gd * (Sigma %*% gd))))
    data.frame(Estimate = est, SE = se.est)
  }

rstpm2 documentation built on Oct. 29, 2024, 5:06 p.m.