Standard Error Estimates through Bootstrap

Usage

bootSe(
  object,
  B = 50,
  se = c("inter-quartile", "mad", "sd"),
  return_beta = FALSE,
  ...
)

Arguments

object

iCoxph-class object.

B

A positive integer specifying number of bootstrap samples used for SE estimates. A large number, such as 200, is often needed for a more reliable estimation in practice. If B = 1 is specified, the function will return the covariate coefficient estimates instead of a iCoxph-class object.

se

A character value specifying the way computing SE from bootstrap samples. The default method is based on median absolute deviation and the second method is based on inter-quartile, both of which are based on normality of the bootstrap estimates and provides robust estimates for SE. The third method estimates SE by the standard deviation of the bootstrap estimates.

return_beta

A logical value. If TRUE, the function returns the covariate coefficient estimates from the given number of bootstrap samples, which allows users to split the most computationally intensive step into small pieces that can be computed in a parallel manner. The default value is FALSE.

...

Other arguments for future usage. A warning will be thrown if any invalid argument is specified.

Value

iCoxph-class object or a numeric matrix that contains the covariate coefficient estimates from the given number of bootstrap samples in rows.

Details

Three different methods are available for computing SE from bootstrap samples through argument se. Given the fact that the bootstrap method is computationally intensive, the function returns the coefficient estimates in a matrix from the given number of bootstrap samples when return_beta = TRUE) is specified, which can be used in parallel computing or high performance computing (HPC) cluster. The SE estimates can be further computed based on estimates from bootstrap samples by users on their own. The return_beta = TRUE is implied, when B = 1 is specified.

See also

Examples

## See examples of function 'iCoxph'.