Interface for CoxBoost selection of the optimal number of boosting steps via cross-validation

Determines the number of boosting steps for a survival model fitted by CoxBoost via cross-validation, conforming to the calling convention required by argument complexity in peperr call.

Usage

complexity.mincv.CoxBoost(response, x, full.data, ...)

Arguments

response

a survival object (Surv(time, status)).

x

n*p matrix of covariates.

full.data

data frame containing response and covariates of the full data set.

...

additional arguments passed to cv.CoxBoost call.

Value

Scalar value giving the optimal number of boosting steps.

Details

Function is basically a wrapper around cv.CoxBoost of package CoxBoost. A K-fold cross-validation (default K = 10) is performed to search the optimal number of boosting steps, per default in the interval [0, maxstepno]. Calling peperr, the default arguments of cv.CoxBoost can be changed by passing a named list containing these as argument args.complexity.

Since CoxBoost is only suggested by peperr, install it before calling this function.

See also

peperr, cv.CoxBoost