ncvreg (version 3.8-0)

perm.ncvreg: Permutation fitting for ncvreg

Description

Fits multiple penalized regression models in which the outcome is randomly permuted, thereby allowing estimation of the false inclusion rate.

Usage

perm.ncvreg(X, y, ..., permute=c("outcome", "residuals"), N=10, seed, trace=FALSE)

Arguments

X

The design matrix, without an intercept, as in ncvreg.

y

The response vector, as in ncvreg.

...

Additional arguments to ncvreg.

permute

What to permute. If 'outcome', the response vector, y, is permuted. If 'residuals', the residuals are permuted. This is only available for linear regression (i.e., for family='gaussian'). Note that permuting the residuals may take a long time, as the residuals differ for each value of lambda, so separate permutations are required at every value of lambda. See also permres.

N

The number of permutation replications. Default is 10.

seed

You may set the seed of the random number generator in order to obtain reproducible results.

trace

If set to TRUE, perm.ncvreg will inform the user of its progress by announcing the beginning of each permutation fit. Default is FALSE.

Value

An object with S3 class "perm.ncvreg" containing:

Details

The function fits a penalized regression model to the actual data, then repeats the process N times with a permuted version of the response vector. This allows estimation of the expected number of variables included by chance for each value of lambda. The ratio of this expected quantity to the number of selected variables using the actual (non-permuted) response is called the false inclusion rate (FIR).

See Also

ncvreg, plot.fir, fir

Examples

## Linear regression
data(prostate)
X <- as.matrix(prostate[,1:8])
y <- prostate$lpsa
pmfit <- perm.ncvreg(X, y)

par(mfcol=c(2,2))
plot(pmfit)
plot(pmfit, type="EF")
plot(pmfit$fit)

pmfit.r <- perm.ncvreg(X, y, permute='residuals')
plot(pmfit.r, col="red")                          # Permuting residuals is
lines(pmfit$fit$lambda, pmfit$FIR, col="gray60")  # less conservative

## Logistic regression
data(heart)
X <- as.matrix(heart[,1:9])
y <- heart$chd
pmfit <- perm.ncvreg(X, y, family="binomial")

par(mfrow=c(2,2))
plot(pmfit)
plot(pmfit, type="EF")
plot(pmfit$fit)