R/booami_sim.R

In booami: Component-Wise Gradient Boosting after Multiple Imputation

#' Example dataset for 'booami' (Gaussian, MAR)
#'
#' A simulated dataset with predictors \code{X1}...\code{X25} and a continuous
#' outcome \code{y}. Missing values are generated under a MAR mechanism in the
#' predictors (covariates) only; the outcome \code{y} is fully observed (no NAs).
#' The object is a \code{data.frame} and carries attributes describing the
#' data-generating process (true coefficients, informative indices, etc.).
#'
#' @format A data frame with 300 rows and 26 variables:
#' \describe{
#'   \item{X1}{numeric}
#'   \item{X2}{numeric}
#'   \item{X3}{numeric}
#'   \item{X4}{numeric}
#'   \item{X5}{numeric}
#'   \item{X6}{numeric}
#'   \item{X7}{numeric}
#'   \item{X8}{numeric}
#'   \item{X9}{numeric}
#'   \item{X10}{numeric}
#'   \item{X11}{numeric}
#'   \item{X12}{numeric}
#'   \item{X13}{numeric}
#'   \item{X14}{numeric}
#'   \item{X15}{numeric}
#'   \item{X16}{numeric}
#'   \item{X17}{numeric}
#'   \item{X18}{numeric}
#'   \item{X19}{numeric}
#'   \item{X20}{numeric}
#'   \item{X21}{numeric}
#'   \item{X22}{numeric}
#'   \item{X23}{numeric}
#'   \item{X24}{numeric}
#'   \item{X25}{numeric}
#'   \item{y}{numeric outcome (fully observed)}
#' }
#'
#' @details
#' Generated by \code{\link{simulate_booami_data}} with typical settings (see
#' \code{?simulate_booami_data}). The following attributes are attached to
#' \code{booami_sim}:
#' \itemize{
#'   \item \code{"true_beta"}: numeric length-25 vector of true coefficients
#'         (non-zeros in positions 1-5).
#'   \item \code{"informative"}: integer vector \code{1:5}.
#'   \item \code{"type"}: \code{"gaussian"}.
#'   \item \code{"corr_structure"}: \code{"all_ar1"}; \code{"rho"}: 0.3.
#'   \item \code{"intercept"}: 1; \code{"noise_sd"}: 1 (Gaussian; \code{NA} otherwise).
#'   \item \code{"mar_scale"}: \code{TRUE}; \code{"keep_mar_drivers"}: \code{TRUE}.
#' }
#'
#' @examples
#' ## \donttest{
#' utils::data(booami_sim)
#' dim(booami_sim)
#' mean(colSums(is.na(booami_sim)) > 0)  # fraction of columns with any NAs
#' sum(is.na(booami_sim$y))              # should be 0
#' head(attr(booami_sim, "true_beta"))
#' attr(booami_sim, "informative")
#' ## }
#'
#' @seealso \code{\link{simulate_booami_data}},
#'   \code{\link{impu_boost}}, \code{\link{cv_boost_raw}}, \code{\link{cv_boost_imputed}}
#' @keywords datasets
#' @name booami_sim
#' @docType data
NULL