estRidge {eshrink}R Documentation

Estimate Coefficients for Ridge Regression

Description

Computes a vector of regression coefficients for a provided ridge penalty.

Usage

estRidge(lambda, X, y, penalize, XtX = crossprod(X))

Arguments

lambda

ridge penalty factor

X

design matrix for the regression.

y

outcome vector. Unless X contains an intercept column, this should typically be centered.

penalize

vector giving penalty structure. Values must be in [0, 1]. See Details for more information.

XtX

(optional) cross product of the design matrix. If running simulations or other procedure for identical X, providing a pre-computed value can reduce computational cost.

Details

The input penalize is a vector of ridge penalty factors, such that the penalty for covariate j is lambda*penalize[j]. Although its primary purpose is for indicating which variables to penalize (1) and which to not penalize (0), fractional values between 0 and 1 are accepted. Defaults to c(0, rep(1, p-1)), where p is number of columns in X (this penalizes all coefficients but the first).

The design matrix X is assumed to contain only numeric values, so any factors should be coded according to desired contrast (e.g., via model.matrix)

Author(s)

Joshua Keller

See Also

festRidge, mseRidge


[Package eshrink version 0.1.2 Index]