Lasso Partial Ridge
Computes the two-stage estimator Lasso+Partial Ridge.
LPR(x, y, lambda = NULL, fix.lambda = TRUE, lambda2, cv.method = "cv", nfolds = 10,
foldid, cv.OLS = TRUE, tau = 0, parallel = FALSE, standardize = TRUE, intercept =
TRUE, ...)x |
Input matrix as in glmnet, of dimension nobs x nvars; each row is an observation vector. |
y |
Response variable. |
lambda |
lambda: A value of lambda - default is NULL. lambda should be given a value when fix.lambda=TRUE. |
fix.lambda |
If TRUE, computes Lasso+Partial Ridge estimator for a fix value of lambda given by the argument "lambda"; otherwise, computes Lasso+Partial Ridge estimator for the value of lambda choosing by cv/cv1se/escv. |
lambda2 |
Tuning parameter in the Partial Ridge. If missing, lambda2 will be set to 1/nobs, where nobs is the number of observations. |
cv.method |
The method used to select lambda – can be cv, cv1se, and escv; the default is cv. cv.method is useful only when fix.lambda=FALSE. |
nfolds, foldid, cv.OLS, tau, parallel |
Arguments that can be passed to escv.glmnet (useful only when fix.lambda=FALSE). Note that, the default value of cv.OLS is TRUE, which means using Lasso+OLS in the cv fits. |
standardize |
Logical flag for x variable standardization, prior to fitting the model. Default is standardize=TRUE. |
intercept |
Should intercept be fitted (default is TRUE) or set to zero (FALSE). |
... |
Other arguments that can be passed to glmnet. |
This function computes the Lasso+Partial Ridge estimator for a give value of lambda (if fix.lambda=TRUE) or for the value of lambda choosing by cv/cv1se/escv (if fix.lambda=FALSE).
A list consisting of the following elements is returned.
beta |
The Lasso+Partial Ridge estimator for the coefficients of variables/predictors. |
beta0 |
A value of intercept term. |
lambda |
The value/values of lambda. |
lambda2 |
The value of lambda2. |
meanx |
The mean vector of variables/predictors if intercept=TRUE, otherwise is a vector of 0's. |
mu |
The mean of the response if intercept=TRUE, otherwise is 0. |
normx |
The vector of standard error of variables/predictors if standardize=TRUE, otherwise is a vector of 1's. |
tau |
Tuning parameter in modified Least Squares (mls). |
library("glmnet")
library("mvtnorm")
## generate the data
set.seed(2015)
n <- 200 # number of obs
p <- 500
s <- 10
beta <- rep(0, p)
beta[1:s] <- runif(s, 1/3, 1)
x <- rmvnorm(n = n, mean = rep(0, p), method = "svd")
signal <- sqrt(mean((x %*% beta)^2))
sigma <- as.numeric(signal / sqrt(10)) # SNR=10
y <- x %*% beta + rnorm(n)
## Lasso+Partial Ridge estimator
# for a given value of lambda
set.seed(0)
obj.escv <- escv.glmnet(x, y)
obj <- LPR(x, y, lambda = obj.escv$lambda.cv)
# Lasso+OLS estimate of the regression coefficients
obj$beta
# intercept term
obj$beta0
# prediction
mypredict(obj, newx = matrix(rnorm(10*p), 10, p))
# for lambda choosing by cross-validation (cv) which uses Lasso+OLS in the cv fit
set.seed(0)
obj <- LPR(x, y, fix.lambda = FALSE)
# for lambda choosing by cross-validation (cv) which uses Lasso in the cv fit
set.seed(0)
obj <- LPR(x, y, fix.lambda = FALSE, cv.OLS = FALSE)Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.