penaltyExtract: Extract the penalty from a lvm object

In bozenne/lava.penalty: Penalized Latent Variable Models

Description

Extract the penalty from a lvm object

Usage

penalty(x, ...)

## S3 method for class 'lvm'
penalty(x, type = NULL, nuclear = FALSE,
  lambdaPerCoef = FALSE, add.names = TRUE, ...)

## S3 method for class 'lvmfit'
penalty(x, type, ...)

## S3 method for class 'penaltyL12'
penalty(x, type = NULL, no.lasso = FALSE,
  no.ridge = FALSE, no.group = FALSE, index.group = NULL, ...)

## S3 method for class 'penaltyNuclear'
penalty(x, type, group = NULL, keep.list = FALSE,
  ...)

Arguments

x	a lvm model
type	the information about the penalty to be returned.
nuclear	should informations be extracted for the nuclear penalty
lambdaPerCoef	when a regularization parameter is extracted should it expanded to match the parameters of the LVM
add.names	when a regularization parameter is extracted should the name of the parameters be added in the output. Only active if lambdaPerCoef is TRUE.
group	the group of penalty parameter to be returned
keep.list	should the format of the results be always a list

Examples

set.seed(10)
n <- 300
formula.lvm <- as.formula(paste0("Y~",paste(paste0("X",1:12), collapse = "+")))
mSim <- lvm(formula.lvm)
df.data <- sim(mSim,n)

pm <- penalize(mSim, c("Y~X1","Y~X4", "Y~X10"))
penalty(pm, type = "link")
penalty(pm, type = c("link","group"))

pm <- penalize(pm, value = paste0("Y~X",5:6), group = 1, add = TRUE)
penalty(pm, type = "link") # all parameters
penalty(pm, type = "link", group = 0) # individually penalized parameters
penalty(pm, type = "link", group = 1) # group 1 penalization
penalty(pm, type = "V")
penalty(pm, type = "V", group = 0)

penalty(pm, type = "lambda1")
penalty(pm, type = "lambda1", lambdaPerCoef = TRUE)
penalty(pm, type = "lambda1", lambdaPerCoef = TRUE, add.names = FALSE)
penalty(pm, type = c("lambda1","lambda2"), lambdaPerCoef = TRUE)

penalty(pm, type = "link") <- c("Y~X1" , "Y~X4")

bozenne/lava.penalty documentation built on May 13, 2019, 1:41 a.m.