An example of TF-IDF

TFIDF.R

library(janeaustenr)
library(tidytext)
library(dplyr)
library(ggplot2)

book_words <- austen_books() %>%
  unnest_tokens(word, text) %>%
  count(book, word, sort = TRUE) %>% 
  group_by(book) %>% 
  mutate(total = sum(n)) %>%
  ungroup() %>% 
  bind_tf_idf(term = word, document = book, n = n) %>% 
  arrange(desc(tf_idf))

book_tf_idf = mutate(book_words, tf2 = n/total) %>% 
  group_by(word) %>% 
  mutate(idf2 = -log(n_distinct(book))) %>% 
  ungroup() %>% 
  mutate(idf2 = idf2 +log(n_distinct(book))) %>% 
  mutate(tf_idf2 = idf2*tf2)

## check
ggplot(book_tf_idf2, aes(tf_idf2 , tf_idf))+
  geom_abline(slope = 1, intercept = 0, linetype=2)+
  geom_point(alpha=0.2)+
  theme_classic(14)

book_top <- book_tf_idf %>%
  group_by(book) %>%
  slice_max(tf_idf, n = 10) %>%
  ungroup()

ggplot(book_top, aes(x = tf_idf, y = fct_reorder(word, tf_idf))) +
  geom_point() +
  geom_segment( aes(xend=0, yend=fct_reorder(word, tf_idf)))+
  facet_wrap(~book, ncol = 2, scales = "free") +
  labs(x = "tf-idf", y ="word") +
  theme_classic(14)