tf_idf_analysis.R

#step 1 : remove rows with merge, pull, request words as they useless for analysis
custom_stop_words<-data.frame("word"=c("merge", "pull", "request"))
tidy_commits<-tidy_commits %>% anti_join(custom_stop_words)

#step 2: calculate tf-idf and sort the rows by its value in descending order  
commit_words<-tidy_commits %>% 
  count(name, word, sort=TRUE) %>% 
  ungroup() %>% 
  bind_tf_idf(word, name, n) %>%
  arrange(desc(tf_idf)) %>%
  mutate(word = reorder(word, tf_idf))


#step 3 : plot the bar chart 
commit_words %>% 
  filter(name=="react") %>% top_n(20) %>% ggplot(aes(word, tf_idf)) +
  geom_bar(stat = "identity", fill=I("#53d2fa")) +
  labs(y="tf-idf", title="Highest tf-idf words in React Commit Messages") +
  coord_flip()