weights.enetLTS: binary weights from the '"enetLTS"' object

In enetLTS: Robust and Sparse Methods for High Dimensional Linear and Binary and Multinomial Regression

binary weights from the "enetLTS" object

Description

Extract binary weights that indicate outliers from the current model.

Usage

## S3 method for class 'enetLTS'
weights(object,vers=c("reweighted","raw","both"),index=FALSE,...)

Arguments

object	the model fit from which to extract outlier weights.
vers	a character string specifying for which estimator to extract outlier weights. Possible values are "reweighted" (the default) for weights indicating outliers from the reweighted fit, "raw" for weights indicating outliers from the raw fit, or "both" for the outlier weights from both estimators.
index	a logical indicating whether the indices of the weight vector should be included or not (the default is FALSE).
...	additional arguments from the enetLTS object if needed.

Value

A numeric vector containing the requested outlier weights.

Note

The weights are 1 for observations with reasonably small residuals and 0 for observations with large residuals. Here, residuals represent standardized residuals for family="gaussian", Pearson residuals for family="binomial" and group-wise scaled robust distances family="multinomial".

Use weights with or without index is available.

Author(s)

Fatma Sevinc KURNAZ, Irene HOFFMANN, Peter FILZMOSER
Maintainer: Fatma Sevinc KURNAZ <[e-mail]>;<[e-mail]>

See Also

enetLTS

Examples

## for gaussian

set.seed(86)
n <- 100; p <- 25                             # number of observations and variables
beta <- rep(0,p); beta[1:6] <- 1              # 10% nonzero coefficients
sigma <- 0.5                                  # controls signal-to-noise ratio
x <- matrix(rnorm(n*p, sigma),nrow=n)
e <- rnorm(n,0,1)                             # error terms
eps <- 0.1                                    # contamination level
m <- ceiling(eps*n)                           # observations to be contaminated
eout <- e; eout[1:m] <- eout[1:m] + 10        # vertical outliers
yout <- c(x %*% beta + sigma * eout)        # response
xout <- x; xout[1:m,] <- xout[1:m,] + 10      # bad leverage points


fit1 <- enetLTS(xout,yout)
weights(fit1)
weights(fit1,vers="raw",index=TRUE)
weights(fit1,vers="both",index=TRUE)


## for binomial

eps <-0.05                                     # %10 contamination to only class 0
m <- ceiling(eps*n)
y <- sample(0:1,n,replace=TRUE)
xout <- x
xout[y==0,][1:m,] <- xout[1:m,] + 10;          # class 0
yout <- y                                      # wrong classification for vertical outliers


fit2 <- enetLTS(xout,yout,family="binomial")
weights(fit2)
weights(fit2,vers="raw",index=TRUE)
weights(fit2,vers="both",index=TRUE)


## for multinomial

n <- 120; p <- 15
NC <- 3
X <- matrix(rnorm(n * p), n, p)
betas <- matrix(1:NC, ncol=NC, nrow=p, byrow=TRUE)
betas[(p-5):p,]=0; betas <- rbind(rep(0,NC),betas)
lv <- cbind(1,X) %*% betas
probs <- exp(lv)/apply(exp(lv),1,sum)
y <- apply(probs,1,function(prob){sample(1:NC, 1, TRUE, prob)})
xout <- X
eps <-0.05                          # %10 contamination to only class 0
m <- ceiling(eps*n)
xout[1:m,] <- xout[1:m,] + 10       # bad leverage points
yout <- y


fit3    <- enetLTS(xout,yout,family="multinomial")
weights(fit3)
weights(fit3,vers="raw",index=TRUE)
weights(fit3,vers="both",index=TRUE)

enetLTS documentation built on May 22, 2022, 1:05 a.m.