R/sim_dfm.R

In contefranz/OpTop: Optimal topic specification for latent dirichlet allocation models

#' Simulate a document-feature-matrix from a LDA specification
#'
#' Simulate a document-feature-matrix from estimated DTW and TWW. The number of topics as well as 
#' the number of documents are inferred from the LDA parameters.
#'
#' @param DTW A matrix or data.frame with Document-Topic-Weights.
#' @param TWW A matrix or data.frame with Topic-Word-Weights.
#' @param doc_length A vector containing the desired document length as total number of word counts.
#' @param alpha Parameter of the Dirichlet distribution for topics over documents.
#' @param seed Input to \code{set.seed}.
#' 
#' @return A \code{\link[quanteda]{dfm}} object.
#' 
#' @seealso \code{\link[topicmodels]{LDA}} \code{\link[quanteda]{dfm}}
#' @references Lewis, C. and Grossetti, F. (2019 - forthcoming):\cr
#' A Statistical Approach for Optimal Topic Model Identification.
#' @author Francesco Grossetti \email{francesco.grossetti@@unibocconi.it}
#' @author Craig M. Lewis \email{craig.lewis@@owen.vanderbilt.edu}
#' @import data.table ggplot2
#' @importFrom LDATS sim_LDA_data
#' @importFrom quanteda as.dfm
#' @export

sim_dfm <- function( DTW, TWW, doc_length, alpha = NULL, seed = NULL) {
  
  if ( all( !is.matrix( DTW ), !is.data.frame( DTW ) ) ) {
    stop( "DTW must be either a matrix or a data.frame" )
  }
  if ( all( !is.matrix( TWW ), !is.data.frame( TWW ) ) ) {
    stop( "TWW must be either a matrix or a data.frame" )
  }
  if ( all( !is.integer( doc_length ), !is.numeric( doc_length ) ) ) {
    stop( "doc_lenght must be either an integer or a numeric vector" )
  }
  
  out <- sim_LDA_data( N = doc_length,
                       Beta = TWW,
                       alpha = alpha,
                       Theta = DTW,
                       seed = seed )
  
  outdfm <- as.dfm( out )
  return( outdfm )
  
}