R/make_analysis_df.R

In Amherst-Statistics/katherinemansfieldr: A Package for Text Mining Katherine Mansfield's Short Stories

#' Create analysis table
#'
#' Takes a collection of chapters, breaks
#' them up into individual chapters, and returns summary dataset
#'for each chapter. Output statistics include average sentence length,
#' average token length, average type length, average variety 
#' (# type/# token), story length (number of words), number of occurences of words (expressed as
#' ratio of frequency of word over total number of words), 
#' number of occurances of punctuation (expressed as ratio of frequency of
#' punctuation mark over all punctuation marks) and average sentiment score
#' (AFINN method).
#'
#' @param text Character vector containing all the lines in a given text 
#' @param book Name of book or story collection from which chapters/stories are taken
#' @param chapters Line index numbers associated with the first line/title 
#'        line of each chapter
#' @param titles Title of each story or chapter in output dataset
#' @param freqwords Words whose frequency of appearance will be recorded
#' @param punctlist Punctuation characters whose frequency of appearance 
#'        will be recorded 
#' @importFrom dplyr %>%
#' @importFrom utils tail
#' @importFrom syuzhet get_sentiment
#' @export
#' @examples
#' sample <- gardenParty[1:452]
#' collection <- "The Garden Party"
#' breaks <- find_chapters(sample, " and other stories, by Katherine Mansfield : ")
#' stories <- c("At the Bay", "Her First Ball", "The Singing Lesson")
#' words <- c("the", "and", "a")
#' punctuation <- c(",", ".", "...")
#' make_analysis_df(text = sample, book = collection, chapters = breaks, titles = stories,
#'                  freqwords = words, punctlist = punctuation)

make_analysis_df <- function(text, book, chapters, titles, freqwords, punctlist){
  collection_name <- c()
  story_title <- c()
  outputdf <- NULL
  for(i in 1:(length(chapters))){
    col_name <- book
    collection_name <- c(collection_name, col_name)
    start <- chapters[i]
    if(i < length(chapters)){
      end <- chapters[i+1] - 1
    }
    if(i == length(chapters)){
      end <- length(text)
    }
    text.lines <- text[start:end]
    story_title <- c(story_title, titles[i])
    subtext <- collapse_text(text.lines)
    
    # average token length
    text.token <- extract_token(subtext)
    text.token.length <- word_length(text.token)
    text.token.averageLength <- mean(text.token.length)
    
    # average type length
    text.type <- extract_type(subtext)
    text.type.length <- word_length(text.type)
    text.type.averageLength <- mean(text.type.length)
    
    # average sentence length 
    text.sentences <- extract_sentences(subtext)
    text.sentences.length <- sentence_length(text.sentences)
    text.sentences.averageLength <- mean(text.sentences.length)
    
    # variety of words (type/token)
    text.variety <- get_variety(subtext)
    text.variety <- mean(text.variety)
    
    statnames <- c("avg_token_length", "avg_type_length", "variety", 
                   "avg_sentence_length", "story_length")
    wordstats <- c(text.token.averageLength, text.type.averageLength, text.variety, 
                   text.sentences.averageLength, length(text.token.length))
    worddf <- data.frame(statnames, wordstats)
    worddf <- tidyr::spread(worddf, statnames, wordstats)
    
    # frequency of most common words
    WordFreq <- charfreq(text.token, freqwords, punctuation = FALSE)
    WordFreq$freq <- WordFreq$freq/length(text.token)
    WordFreq <- tidyr::spread(WordFreq, character, freq)
    
    # punctuation
    text.punct <- extract_punct(subtext)
    PunctFreq <- charfreq(text.punct, punctlist, punctuation = TRUE)
    PunctFreq$freq <- PunctFreq$freq/length(text.punct)
    PunctFreq <- tidyr::spread(PunctFreq, character, freq)
    
    text.sentiment <- extract_sentences(subtext)
    sentiment <- syuzhet::get_sentiment(text.sentiment, method = "afinn")
    sentmean <- tail(sentiment, 100)%>%
      mean()
    
    chapterSummary <- cbind(worddf, WordFreq, PunctFreq, sentmean)
    if(i == 1){
      outputdf <- chapterSummary
    }
    else{
      outputdf <- rbind(outputdf, chapterSummary)
    }
  }
  
  outputdf <- cbind(collection_name, story_title, outputdf)
  return(outputdf)
}