survsrf_tune {survcompare}

Internal function to tune SRF model, in nested CV loop

Description

Internal function to tune SRF model, in nested CV loop

Usage

survsrf_tune(
  df_tune,
  predict.factors,
  inner_cv = 3,
  fixed_time = NaN,
  randomseed = NULL,
  mtry = c(3, 4, 5),
  nodesize = c(10, 20, 50),
  nodedepth = c(100),
  verbose = FALSE,
  oob = TRUE
)

Arguments

df_tune

data frame

predict.factors

predictor names

inner_cv

k in k-fold CV, applied if oob=FALSE

fixed_time

NaN

randomseed

random seed

mtry

tuning parameter

nodesize

at which event probabilities are computed

nodedepth

tuning parameter

verbose

FALSE

oob

TRUE/FALSE use out-of-bag predictions while tuning instead of cross-validation, default is TRUE and is faster

Value

output=list(modelstats, bestbrier, bestauc, bestcindex)