Implements covariance-regularized regression, aka the Scout Method.

Description

Functions for implementing covariance-regularize regression.

Details

Package: scout
Type: Package
Version: 1.0
Date: 2008-11-20
License: GPL (>=2)

The main function is "scout", which takes in a data matrix x and an outcome vector y and estimates regression coefficients for Scout(2,1) for a range of tuning parameter values. Alternatively one can specify other tuning parameter values and one can also perform Scout(1,1), Scout(2,.), or Scout(1,.). Cross-validation and prediction functions also are available.

Author(s)

Daniela Witten and Robert Tibshirani

Maintainer: Daniela Witten <dwitten@stanford.edu>

References

Witten and Tibshirani (2008) Covariance-regularized regression and classification for high-dimensional problems. Journal of the Royal Statistical Society, Series B 71(3): 615-636.

See Also

<http://www-stat.stanford.edu/~dwitten>

Examples

1
2
3
4
5
6
7
8
library(lars)
data(diabetes)
attach(diabetes)
## Not run: cv.out <- cv.scout(x2,y,p1=1,p2=1,K=3)
## Not run: print(cv.out)
## Not run: out <- scout(x2,y,p1=1,p2=1,lam1=cv.out$bestlam1,lam2=cv.out$bestlam2)
## Not run: coef <- out$coef[1,1,]
detach(diabetes)