sprinter (version 1.1.0)

resample.sprinter: Wrapper for the function sprinter in order to judge the stability of the selected variables

Description

resample.sprinter is a wrapper, which subsamples the data and performs the function sprinter on each subsample. It evaluates the interaction terms by an improvement in prediction performance. The importance of an interaction is judged by variable inclusion frequencies and mean coefficients for each interaction term.

Usage

resample.sprinter(x, time, 
                  status,
                  fold = 5,
                  oob.rel = 0.632,
                  mandatory,
                  repetitions = 25, 
                  n.inter.candidates = 1000, 
                  screen.main, 
                  screen.inter = fit.rf, 
                  fit.final,
                  args.screen.main = list(), 
                  args.screen.inter = list(),
                  args.fit.final = args.screen.main, 
                  orthogonalize = TRUE,
                  parallel = FALSE, 
                  mc.cores = detectCores(), ...)

Arguments

x

n * p matrix of covariates.

time

vector of length n specifying the observed times.

status

censoring indicator, i.e., vector of length n with entries 0 for censored observations and 1 for uncensored observations.

fold

number of subsamples. Default is 5.

oob.rel

frequency of subsamples.

mandatory

vector with variable names of mandatory covariates, where parameter estimation should be performed unpenalized.

repetitions

number of repetitions of the interaction screening approach. Repetitions are performed by creating subsamples and applying the interaction screening approach on each subsample dataset separately.

n.inter.candidates

minimal number of potential interaction candidates, which are considered in the final model building step.

screen.main

function for screening potential main effects. See more in details.

screen.inter

function for detecting potential interaction candidates. See more in details.

fit.final

function for building final Cox proportional hazards model. Default is the function set in screen.main. See more in details.

args.screen.main

list of arguments which should be used in the main effects detection step.

args.screen.inter

list of arguments which should be used in the interaction screening step.

args.fit.final

list of arguments which should be used in the final model building step.

orthogonalize

logical value. If true all variables are made orthogonal to those that are assessed as main effects by screen.main.

parallel

logical value indicating whether the interaction screening step should be performed parallel.

mc.cores

the numbers of cores to use.

...

additional arguments.

Value

• Returns a list of length fold of objects produced by sprinter.

Details

resample.sprinter provides the possibility of evaluating the relevance of an interaction by using resampling techniques and the resultant variable inclusion frequency. This function subsamples the data and applies the function sprinter on each subsample. As a result the user can print the interactions with the largest inclusion frequencies and their mean coefficients.

References

Sariyar, M., Hoffmann, I. Binder, H. (2014). Combining techniques for screening and evaluating interaction terms on high-dimensional time-to-event data. BMC Bioinformatics 15:58.

See Also

sprinter

Examples

############
# Example for resampling and screening main effects by CoxBoost, 
# screening interactions by random forest and 
# building a final Cox proportional hazards model by CoxBoost:
# system.time with one core:
#   user  system elapsed 
# 399.78    2.40  403.02 
# For a faster run set repetitions down!
############

# Create survival data with interactions:
data <- simul.int(518,n = 200, p = 500,
                   beta.int = 1.0,
                   beta.main = 0.9, 
                   censparam = 1/20, 
                   lambda = 1/20)[[1]]
set.seed(123)

resamcb <- resample.sprinter(  x=data[,1:500],  
                               time = data$obs.time,
                               status= data$obs.status,
                               fold = 10,
                               repetitions = 5, 
                               mandatory = c("ID1","ID2"),
                               n.inter.candidates = 1000, 
                               screen.main = fit.CoxBoost, 
                               fit.final = fit.CoxBoost,  
                               args.screen.main = list(stepno = 5),
                               parallel = F)
summary(resamcb)
summary(resamcb, optional = T)
# true coefficients:
# Clin.cov1   Clin.cov2   ID5:ID6   ID7:ID8
# 0.9         -0.9        1         -1


## Example for resampling and screening main effects by CoxBoost, 
## screening interactions by random forest and 
## building a final Cox proportional hazards model by CoxBoost:
##
##system.time with one core:
##   user  system elapsed 
##  479.55    2.39  482.40  
## For a faster run set repetitions down!

# Create survival data with interactions:
data <- simul.int(518,n = 200, p = 500,
                   beta.int = 1.0,
                   beta.main = 0.9, 
                   censparam = 1/20, 
                   lambda = 1/20)[[1]]

resamunicox <- resample.sprinter(  x=data[,1:500],  
                               time = data$obs.time,
                               status= data$obs.status,
                               fold = 10,
                               repetitions = 10, 
                               mandatory = c("ID1","ID2"),
                               n.inter.candidates = 1000, 
                               screen.main = fit.uniCox, 
                               fit.final = fit.uniCox,  
                               parallel = FALSE)
                               
                               
summary(resamunicox)
summary(resamunicox, optional = TRUE)