R/get_word_counts.R

In matthewjrogers/mrtools: Matthew J. Rogers's convenience functions

#' Generate word counts
#'
#' @param data A data.frame
#' @param column A column name as a string
#' @param exclude_words An optional vector of words to exclude from the counts. Stop words (e.g. and, but, or) are handled automatically.
#' @param min_count The minimum word frequency to include in the final counts. Useful for pruning wordclouds.
#'
#' @return A data.frame of word counts, optionally by sentiment
#' @export
#' @import tidytext
#' @import dplyr
#'
tokenize_column <- function(data, column, exclude_words = '', min_count = 0){
  tokenized_data <- data %>%
    dplyr::filter(!is.na(!!as.name(column))) %>%
    dplyr::select(!!as.name(column)) %>%
    tidytext::unnest_tokens(.data$word, !!as.name(column)) %>%
    dplyr::anti_join(tidytext::stop_words, by = 'word') %>%
    dplyr::group_by(.data$word) %>%
    dplyr::filter(!stringr::str_detect(.data$word, "^\\d"),
                  !.data$word %in% exclude_words
    ) %>%
    dplyr::count() %>%
    dplyr::filter(.data$n > min_count) %>%
    dplyr::arrange(desc(.data$n))

  return(tokenized_data)
}
# get_word_counts <- function(df, column, sentiments = NULL, exclude_words = "", min_count = 1){
#
#   if (!is.null(sentiments) & !is.data.frame(sentiments)){
#     stop("Sentiments must be a data.frame from tidytext::get_sentiments('bing')")
#   }
#
#   tokens <- df %>%
#     tidytext::unnest_tokens(.data$word, !!as.name(column)) %>%
#     dplyr::anti_join(tidytext::stop_words, by = "word") %>%
#     dplyr::filter(!is.na(.data$word),
#                   !.data$word %in% exclude_words
#                   )
#
#   if (!is.null(sentiments)){
#
#     tokens <- dplyr::inner_join(tokens, sentiments, by = "word") %>%
#       dplyr::group_by(.data$word, .data$sentiment) %>%
#       dplyr::count() %>%
#       dplyr::filter(.data$n >= min_count) %>%
#       dplyr::arrange(.data$sentiment, -.data$n)
#
#   } else{
#
#     tokens <- tokens %>%
#       dplyr::group_by(.data$word) %>%
#       dplyr::count() %>%
#       dplyr::filter(.data$n >= min_count) %>%
#       dplyr::arrange(-.data$n)
#
#   }
#
#   return(tokens)
#
# }