R/model-training-plan.R

In mdneuzerling/DrakeModelling: Using drake to Train and Execute a Model in a Custom Package

#' drake plan for model training
#'
#' @return A plan to be run with drake::make()
#' @importFrom drake trigger
#' @export
#'
model_training_plan <- function() {
  drake::drake_plan(
    reviews = target(
      download_and_read_data(
        file_in("https://archive.ics.uci.edu/ml/machine-learning-databases/00331/sentiment%20labelled%20sentences.zip")
      )
    ),
    vocabulary = create_vocabulary(reviews$review,
                                   doc_proportion_min = 25 / nrow(reviews)),
    vectoriser = text2vec::vocab_vectorizer(vocabulary),
    dtm_unweighted = map_to_dtm(reviews$review,
                                vectoriser = vectoriser),
    tfidf = create_tfidf(dtm_unweighted),
    dtm_tfidf_weighted = map_to_dtm(reviews$review,
                                    vectoriser = vectoriser,
                                    tfidf = tfidf),
    review_rf = randomForest::randomForest(
      x = as.matrix(dtm_tfidf_weighted),
      y = factor(reviews$sentiment),
      ntree = 500
    ),
    validation = validate_model(review_rf, vectoriser, tfidf),
    output_model = drake::target(
      {
        dir.create("artefacts", showWarnings = FALSE)
        readr::write_rds(vectoriser, file_out("artefacts/vectoriser.rds"))
        readr::write_rds(tfidf, file_out("artefacts/tfidf.rds"))
        readr::write_rds(review_rf, file_out("artefacts/review_rf.rds"))
      },
      trigger = drake::trigger(condition = validation, mode = "blacklist")
    )
  )
}