Description Details Author(s) References See Also Examples
Robust dimension reduction methods for regression and discriminant analysis are implemented that yield estimates with a partial least squares alike interpretability. Partial robust M regression (PRM) is robust to both vertical outliers and leverage points. Sparse partial robust M regression (SPRM) is a related robust method with sparse coefficient estimate, and therefore with intrinsic variable selection. For bianry classification related discriminant methods are PRM-DA and SPRM-DA.
Package: | sprm |
Type: | Package |
Version: | 1.1 |
Date: | 2014-12-10 |
License: | GPL(>=3) |
The main functions in this package are prms
and sprms
for non-spares and sparse partial robust M regression, respectively, and prmda
and sprmda
for non-spares and sparse partial robust M discriminant analysis. Further cross validation procedures for tuning parameter selection are implemented in prmsCV
, sprmsCV
, prmdaCV
and sprmdaCV
. Various plot options are available to visualize the results.
Sven Serneels, BASF Corp and Irene Hoffmann
Hoffmann, I., Filzmoser, P., Serneels, S., Varmuza, K., Sparse and robust PLS for binary classification. In print.
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.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 | set.seed(50235)
U1 <- c(rep(3,20), rep(4,30))
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
mod <- prms(y~., data=d, a=2, fun="Hampel")
smod <- sprms(y~., data=d, a=2, eta=0.5, fun="Hampel")
biplot(mod)
biplot(smod)
|
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.