Predict method for models of class sprm

Description

Predictions from a sparse partial robust M regression model.

Usage

1
2
## S3 method for class 'sprm'
predict(object, newdata, ...)

Arguments

object

object of class sprm.

newdata

optional data frame with new observations.

...

further arguments. Currently not used.

Details

If newdata is specified the sprm model is used to predict the fitted values for this data set, otherwise the fitted values of the model are returned.

Value

predict.sprm returns a vector of the predicted response.

Author(s)

Sven Serneels, BASF Corp and Irene Hoffmann

References

Hoffmann, I., Serneels, S., Filzmoser, P., Croux, C. (2015). Sparse partial robust M regression. Chemometrics and Intelligent Laboratory Systems, 149, 50-59.

Serneels, S., Croux, C., Filzmoser, P., Van Espen, P.J. (2005). Partial Robust M-Regression. Chemometrics and Intelligent Laboratory Systems, 79, 55-64.

See Also

sprms, sprmsCV

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
set.seed(5023)
U1 <- c(rep(2,20), rep(5,30))
U2 <- rep(3.5,50)
X1 <- replicate(5, U1+rnorm(50))
X2 <- replicate(20, U2+rnorm(50))
X <- cbind(X1,X2)
beta <- c(rep(1, 5), rep(0,20))
e <- c(rnorm(45,0,1.5),rnorm(5,-20,1))
y <- X%*%beta + e
d <- as.data.frame(X)
d$y <- y
smod <- sprms(y~., data=d, a=1, eta=0.5, fun="Hampel")

dnew <- as.data.frame(cbind(replicate(5, U1+rnorm(10)), replicate(20, U2+rnorm(10))))
ynewp <- predict(smod, newdata=dnew)