R/posneg_words.R

In evan-l-munson/SAoTD: Sentiment Analysis of Twitter Data

#' @title Twitter Positive and Negative Words
#'
#' @description Determines and displays the most positive and negative words 
#'   within the twitter data.
#'
#' @param DataFrameTidy DataFrame of Twitter Data that has been tidy'd.
#' @param num_words Desired number of words to be returned.
#' @param filterword Word or words to be removed.
#'
#' @importFrom dplyr mutate inner_join group_by count filter ungroup top_n quo
#' @importFrom tidytext get_sentiments
#' @importFrom stats reorder
#' @import ggplot2
#'
#' @return A ggplot
#'
#' @examples
#' \dontrun{
#' library(saotd)
#' data <- raw_tweets
#' tidy_data <- Tidy(DataFrame = data)
#' posneg <- posneg_words(DataFrameTidy = tidy_data,
#'                        n = 10)
#' posneg
#'
#' data <- raw_tweets
#' tidy_data <- Tidy(DataFrame = data)
#' posneg <- posneg_words(DataFrameTidy = tidy_data,
#'                        n = 10,
#'                        filterword = "fail")
#' posneg
#'
#' data <- raw_tweets
#' tidy_data <- Tidy(DataFrame = data)
#' posneg <- posneg_words(DataFrameTidy = tidy_data,
#'                        n = 10,
#'                        filterword = c("fail", "urgent"))            
#' posneg
#' }
#' @export

posneg_words <- function(DataFrameTidy,
                         num_words,
                         filterword = NULL) {

  # input checks
  if (!is.data.frame(DataFrameTidy)) {
    stop("The input for this function is a data frame.")
  }

  if (!is.numeric(num_words)) {
    stop("Enter a number.")
  }

  # configure defusing operators for packages checking
  Token <- dplyr::quo(Token)
  sentiment <- dplyr::quo(sentiment)
  n <- dplyr::quo(n)

  # function main body
  TD_PosNeg_Words <- DataFrameTidy %>%
    dplyr::inner_join(
      y = tidytext::get_sentiments(lexicon = "bing"),
      by = c("Token" = "word")) %>%
    dplyr::filter(!(Token %in% filterword)) %>%
    dplyr::count(Token, sentiment) %>%
    dplyr::ungroup() %>%
    dplyr::group_by(sentiment) %>%
    dplyr::top_n(n = num_words) %>%
    dplyr::ungroup() %>%
    dplyr::mutate(Token = stats::reorder(Token, n)) %>%
    ggplot2::ggplot(ggplot2::aes(Token, n, fill = sentiment)) +
    ggplot2::geom_col(show.legend = FALSE) +
    ggplot2::facet_wrap(~sentiment, scales = "free_y") +
    ggplot2::labs(y = "Count",
                  x = NULL) +
    ggplot2::theme_bw() +
    ggplot2::ggtitle("Most common positive and negative words utilizing the Bing Lexicon") +
    ggplot2::coord_flip()

  return(TD_PosNeg_Words)

}