R/methods.R

In gaow/mmbr: Multivariate Sum of Single Effects Regression

#' @title Extract Coefficient Estimates from mvsusie Fit
#'
#' @param object An mvsusie fit, such as the output from a call to
#'   \code{\link{mvsusie}} or \code{\link{mvsusie_rss}}.
#'
#' @param \dots Additional arguments (currently unused).
#'
#' @return An (J+1) x R matrix, where J is the number of predictors
#'   and R is the number of outcomes or response variables. The first
#'   row gives the intercept estimate.
#'
#' @importFrom stats coef
#'
#' @method coef mvsusie
#'
#' @export coef.mvsusie
#'
#' @export
#'
coef.mvsusie <- function(object, ...) {
  return(as.matrix(object$coef))
}

#' @title Predict Outcomes from mvsusie Fit.
#'
#' @param object An mvsusie fit, such as the output from a call to
#'   \code{\link{mvsusie}} or \code{\link{mvsusie_rss}}.
#'
#' @param newx A new X matrix for which to do predictions.
#'
#' @param \dots Additional arguments (currently unused).
#'
#' @return A matrix of predicted outcomes, with rows corresponding to
#'   samples (rows of X), and columns corresponding to outcomes.
#'
#' @importFrom stats coef
#' @importFrom stats predict
#'
#' @method predict mvsusie
#'
#' @export predict.mvsusie
#'
#' @export
#'
predict.mvsusie <- function(object, newx = NULL, ...) {
  if (missing(newx)) {
    return(object$fitted)
  } else {
    b <- coef(object)
    n <- nrow(newx)
    r <- ncol(b)
    intercept <- b[1, ]
    b <- b[-1, ]
    return(matrix(intercept, n, r, byrow = TRUE) + newx %*% b)
  }
}