pseudoCure (version 1.0.0)

geelm: Generalized Estimating Equation with Gaussian family

Description

Fits a generalized estimating equation (GEE) model with Gaussian family with different link functions. The geelm function also supports LASSO or SCAD regularization.

Usage

geelm(
  formula,
  data,
  subset,
  id,
  link = c("identity", "log", "cloglog", "logit"),
  corstr = c("independence", "exchangeable", "ar1"),
  lambda,
  exclude,
  penalty = c("lasso", "scad"),
  nfolds = 5,
  nlambda = 200,
  binit,
  tol = 1e-07,
  maxit = 100
)

Value

An object of class "geelm" representing a linear model fit with GEE.

Arguments

formula

A formula object starting with ~ for the model formula.

data

An optional data frame that contains the covariates and response variables.

subset

An optional logical vector specifying a subset of observations to be used in the fitting process.

id

A vector which identifies the clusters. If not specified, each observation is treated as its own cluster.

link

A character string specifying the model link function. Available options are "identity", "log", "cloglog", and "logit".

corstr

A character string specifying the correlation structure. Available options are "independence", "exchangeable", and "ar1".

lambda

An option for specifying the tuning parameter used in penalization. When this is unspecified or has a NULL value, penalization will not be applied and pCure() will uses all covariates specified in the formulas. Alternatively, this can be specified as a vector numeric vector of non-negative values or "auto" for auto selection.

exclude

A binary numerical vector specifying which variables to exclude in variable selection. The length of exclude must match with the number of covariates. A value of 1 means to exclude in the variable selection.

penalty

A character string specifying the penalty function. The available options are "lasso" and "scad".

nfolds

An optional integer value specifying the number of folds. The default value is 5.

nlambda

An optional integer value specifying the number of tuning parameters to try if lambda = "auto".

binit

A optional numerical vector for the initial value. A zero vector is used when not specified.

tol

A positive numerical value specifying the absolute error tolerance in root search. Default at 1e-7.

maxit

A positive integer specifying the maximum number of iteration. Default at 100.

Examples

gendat <- function() {
  id <- gl(50, 4, 200)
  visit <- rep(1:4, 50)
  x1 <- rbinom(200, 1, 0.6)
  x2 <- runif(200, 0, 1)
  phi <- 1 + 2 * x1
  rhomat <- 0.667^outer(1:4, 1:4, function(x, y) abs(x - y))
  chol.u <- chol(rhomat)
  noise <- as.vector(sapply(1:50, function(x) chol.u %*% rnorm(4)))
  e <- sqrt(phi) * noise
  y <- 1 + 3 * x1 - 2 * x2 + e
  dat <- data.frame(y, id, visit, x1, x2)
  dat
}

set.seed(1); str(dat <- gendat())
geelm(y ~ x1 + x2, id = id, data = dat, corstr = "ar1")