[R] Accept CSR data for predictions (#7615)

This commit is contained in:
david-cortes
2022-01-29 18:54:57 +02:00
committed by GitHub
parent 549bd419bb
commit 7f738e7f6f
8 changed files with 93 additions and 7 deletions

View File

@@ -4,8 +4,10 @@
#' Supported input file formats are either a LIBSVM text file or a binary file that was created previously by
#' \code{\link{xgb.DMatrix.save}}).
#'
#' @param data a \code{matrix} object (either numeric or integer), a \code{dgCMatrix} object, or a character
#' string representing a filename.
#' @param data a \code{matrix} object (either numeric or integer), a \code{dgCMatrix} object,
#' a \code{dgRMatrix} object (only when making predictions from a fitted model),
#' a \code{dsparseVector} object (only when making predictions from a fitted model, will be
#' interpreted as a row vector), or a character string representing a filename.
#' @param info a named list of additional information to store in the \code{xgb.DMatrix} object.
#' See \code{\link{setinfo}} for the specific allowed kinds of
#' @param missing a float value to represents missing values in data (used only when input is a dense matrix).
@@ -37,6 +39,17 @@ xgb.DMatrix <- function(data, info = list(), missing = NA, silent = FALSE, nthre
XGDMatrixCreateFromCSC_R, data@p, data@i, data@x, nrow(data), as.integer(NVL(nthread, -1))
)
cnames <- colnames(data)
} else if (inherits(data, "dgRMatrix")) {
handle <- .Call(
XGDMatrixCreateFromCSR_R, data@p, data@j, data@x, ncol(data), as.integer(NVL(nthread, -1))
)
cnames <- colnames(data)
} else if (inherits(data, "dsparseVector")) {
indptr <- c(0L, as.integer(length(data@i)))
ind <- as.integer(data@i) - 1L
handle <- .Call(
XGDMatrixCreateFromCSR_R, indptr, ind, data@x, length(data), as.integer(NVL(nthread, -1))
)
} else {
stop("xgb.DMatrix does not support construction from ", typeof(data))
}