Bayesian Information Criterion (BIC) — BIC.cox_cure

Compute Bayesian information criterion (BIC) or Schwarz's Bayesian criterion (SBC) from a fitted solution path.

Usage

# S3 method for class 'cox_cure_net'
BIC(object, ..., method = c("obs", "effective"))

# S3 method for class 'cox_cure_net_uncer'
BIC(object, ..., method = c("obs", "effective"))

Arguments

object: An object for a fitted solution path.
...: Other arguments for future usage. A warning message will be thrown for any invalid argument.
method: A character string specifying the method for computing the BIC values. Notice that this argument is placed after ... and thus must be specified as a named argument. The available options for cox_cure objects are "obs" for regular BIC based on the number of observations, and "effective" for using BIC based on the number of effective sample size for censored data (number of uncensored events) proposed by Volinsky and Raftery (2000). The available options for cox_cure_mcar objects are "obs" for regular BIC based on the number of observations, and "certain-event" for a variant of BIC based on the number of certain uncensored events. For objects of either class, the former method is used by default.

References

Volinsky, C. T., & Raftery, A. E. (2000). Bayesian information criterion for censored survival models. Biometrics, 56(1), 256–262.

Examples

## See examples of function 'cox_cure_net'.