carVarSelect: Variable selection with Correlation-Adjusted Regression...

Description Usage Arguments Value Note Author(s) References See Also Examples

Description

Computes the CARS scores and selects significant variables. If the false non discovery rate (fndr) approach is used, significant and null variables are distinguished by an a priori defined q-value.

Usage

1
2
carVarSelect(carSurvScores, method = "fndr", plotDiag = FALSE,
  threshold = 0.05)

Arguments

carSurvScores

Estimated CAR survival scores of each variable (numeric vector). See function carSurvScore.

method

Gives the variable selection procedure. Default is "fndr", which is based on the false non-discovery-rate. The other option is "threshold", which selects only variables above a given empirical quantile.

plotDiag

Should diagnostic plots of the null distribution be plotted? Default is FALSE (logical scalar).

threshold

If method="threshold", then this specifies the quantile threshold of the CAR survival scores. Every score above this threshold is then a significant variable.

Value

Index giving the significant variables of the original data (integer vector).

Note

The quality of estimated, significant variables depends on the sample size and on the number of variables.

Author(s)

Thomas Welchowski

References

Strimmer, K., (2008), A unified approach to false discovery rate estimation, BMC Bioinformatics

See Also

carSurvScore

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
##########################################
# Simulate accelerated, failure time model

# Generate multivariate normal distributed covariates
noObs <- 100
noCovar <- 250
library(mvtnorm)
set.seed(7903)
X <- rmvnorm(noObs, mean=rep(0, noCovar), sigma=diag(noCovar))

# Generate gamma distributed survival times
# Only the first 5 variables have an influence
eta <- 1 - 2 * X[,1] - X[,2] + X[,3] +
0.5 * X[,4] + 1.5 * X[,5]

# Function to generate survival times
genSurv <- function(x) {
set.seed(x)
rgamma(1, shape=2, scale=exp(eta[x]))
}

# Generate survival times
survT <- sapply(1:length(eta), genSurv)

# Generate exponential distributed censoring times
censT <- rexp(noObs, rate=1)

# Calculate event indicator
eventInd <- ifelse(survT <= censT, 1, 0)

# Calculate observed times
obsTime <- survT
obsTime[survT > censT] <- censT [survT > censT]

# Conduct variable selection using fndr
carScores <- carSurvScore(obsTime=obsTime, obsEvent=eventInd, X=X)
selectedVar <- carVarSelect(carSurvScores=carScores)
selectedVar

# Check true positive and true negative rate
TPR <- mean(c(1:5) %in% selectedVar)
TNR <- mean(c(6:250) %in% setdiff(6:250, selectedVar))
perf <- TPR + TNR -1
perf

carSurv documentation built on May 1, 2019, 8:44 p.m.