ridge: Ridge regression
In pbreheny/hdrm: High-Dimensional Regression Modeling

ridge

R Documentation

Ridge regression

Description

Ridge regression

Usage

ridge(obj, ...)

## S3 method for class 'formula'
ridge(obj, data, subset, na.action, contrasts = NULL, ...)

## S3 method for class 'matrix'
ridge(obj, y, lambda, ...)

## S3 method for class 'ridge'
coef(
  object,
  lambda,
  which = 1:length(object$lambda),
  standardize = FALSE,
  drop = TRUE,
  ...
)

## S3 method for class 'ridge'
summary(object, lambda, which, ...)

## S3 method for class 'ridge'
predict(object, X, lambda, which = 1:length(object$lambda), drop = TRUE, ...)

## S3 method for class 'ridge'
confint(object, parm, level = 0.95, X, lambda, which, ...)

Arguments

`obj`	There are two options for running `ridge()`; either supply a formula and data frame as in `lm()`, or an X and y matrix as in `glmnet()`; see examples
`...`	Additional arguments to be passed to methods
`data`	If using the formula interface, a data.frame/list/environment containing the variables in the formula
`subset`	If using the formula interface, a vector describing the subset of the data to be used in fitting the model
`na.action`	If using the formula interface, a function indicating what should happen when the data contain NAs, as in `lm()`
`contrasts`	If using the formula interface, a list to be passed to `model.matrix()`
`y`	If using the matrix interface, the vector of responses
`lambda`	An optional value or vector of values for the regression coefficient; if not supplied, this will be set up automatically
`object`	A `ridge` object, as returned by `ridge()`
`which`	Indices of `lambda` at which coefficients / predictions are required. By default, all indices are returned. If `lambda` is specified, this will override `which`.
`standardize`	Return coefficients on standardized scale (default: FALSE)
`drop`	If requesting coefficients for a single lambda value, drop matrix down to a vector (default: TRUE)
`X`	Matrix of predictor values at which predictions are required.
`parm`	Which parameters to construct confidence intervals for; either a vector of indices or a vector of names (default: all)
`level`	Confidence level (default: 0.95)

Examples

attachData(pollution)
fit <- ridge(X, y)
summary(fit, lambda=0.1)
plot(fit)
plot(fit, xaxis='df')
plot(fit, xaxis='both')
plot(fit, standardize=TRUE)
coef(fit)
coef(fit, standardize=TRUE)
coef(fit, standardize=TRUE, lambda=0.1)
p1 <- cbind(1, std(X)) %*% coef(fit, standardize=TRUE, lambda=1)
p2 <- cbind(1, X) %*% coef(fit, lambda=1)
head(p1); head(p2) # identical

pbreheny/hdrm documentation built on Jan. 17, 2024, 8:53 p.m.