In this project, we will compare bigrams and word frequencies in twitter messages from Trump and Obama. First we will read in the required libraries.

library(dplyr)
library(knitr)
library(purrr)
library(twitteR)
library(tidytext)
library(stringr)
library(wordcloud2)
library(tidyr)
library(ggplot2)
library(igraph)
library(ggraph)

To set up TwitteR, please see R Bloggers.

Now we’ll load the last 3200 tweets from each of Trump and Obama. Note the number of tweets will be less since we are excluding retweets.

trump_tweets <- userTimeline("realDonaldTrump", n=3200)
trump_tweets_df <- tbl_df(map_df(trump_tweets, as.data.frame))

obama_tweets <- userTimeline("BarackObama", n=3200)
obama_tweets_df <- tbl_df(map_df(obama_tweets, as.data.frame))

For Trump

Bigrams

First we’ll look at bigrams contained in Trump’s twitter messages.

count_bigrams <- function(dataset) {
  dataset %>%
    filter(!str_detect(text, '^"')) %>%
    mutate(text = str_replace_all(text, "https://t.co/[A-Za-z\\d]+|&amp;", "")) %>%
    unnest_tokens(bigram, text, token = "ngrams", n = 2) %>%
    separate(bigram, c("word1", "word2"), sep = " ") %>%
    filter(!word1 %in% stop_words$word, !word2 %in% stop_words$word) %>%
    count(word1, word2, sort = TRUE)
}
visualize_bigrams <- function(bigrams) {
  set.seed(5)
  a <- grid::arrow(type = "closed", length = unit(0.1, "inches"))
  bigrams %>%
    graph_from_data_frame() %>%
    ggraph(layout = "fr") +
    geom_edge_link(aes(size = 10, color = "blue"), show.legend = FALSE, arrow = a) +
    geom_node_point(color = "lightblue", size = 16) +
    geom_node_text(aes(label = name, size = 20), repel = TRUE) +
    theme_void() +
    theme(legend.position="none")
}
trump_tweets_df  %>% count_bigrams()  %>% head(50) %>% visualize_bigrams()

Word Frequencies

Now we’ll count word frequencies and create a word cloud.

tweets <- trump_tweets_df

reg <- "([^A-Za-z\\d#@']|'(?![A-Za-z\\d#@]))"
tweet_words <- tweets %>%
  filter(!str_detect(text, '^"')) %>%
  mutate(text = str_replace_all(text, "https://t.co/[A-Za-z\\d]+|&amp;", "")) %>%
  unnest_tokens(word, text, token = "regex", pattern = reg) %>%
  filter(!word %in% stop_words$word,
         str_detect(word, "[a-z]"))

tweet_words <- tweet_words %>% group_by(word) %>% tally() %>% mutate(freq = n) %>% select(word, freq) %>% arrange(desc(freq))
tweet_words %>% head(20) %>% kable(row.names=TRUE)
word freq
1 fake 47
2 news 47
3 media 44
4 people 42
5 jobs 39
6 america 37
7 obamacare 36
8 time 35
9 election 33
10 country 30
11 democrats 30
12 bad 28
13 trump 28
14 american 27
15 healthcare 27
16 join 26
17 president 26
18 russia 25
19 honor 23
20 watch 22 
wordcloud2(tweet_words, size = 0.6)


Sentiment

Now we’ll analyze sentiment.

trump_sentiment <- tweet_words %>%
  inner_join(get_sentiments("bing"))

trump_sentiment %>%
  group_by(sentiment) %>%
  top_n(10, freq) %>%
  arrange(freq) %>%
  ungroup() %>%
  mutate(word = reorder(word, freq)) %>%
  ggplot(aes(word, freq, fill = sentiment)) +
  geom_col(show.legend = FALSE) +
  facet_wrap(~sentiment, scales = "free_y") +
  labs(y = "Contribution to sentiment",
       x = NULL) +
  coord_flip()

For Obama

Bigrams

Now we’ll look at bigrams contained in Obama’s twitter messages.

count_bigrams <- function(dataset) {
  dataset %>%
    filter(!str_detect(text, '^"')) %>%
    mutate(text = str_replace_all(text, "https://t.co/[A-Za-z\\d]+|&amp;", "")) %>%
    unnest_tokens(bigram, text, token = "ngrams", n = 2) %>%
    separate(bigram, c("word1", "word2"), sep = " ") %>%
    filter(!word1 %in% stop_words$word, !word2 %in% stop_words$word) %>%
    count(word1, word2, sort = TRUE)
}
visualize_bigrams <- function(bigrams) {
  set.seed(5)
  a <- grid::arrow(type = "closed", length = unit(0.1, "inches"))
  bigrams %>%
    graph_from_data_frame() %>%
    ggraph(layout = "fr") +
    geom_edge_link(aes(size = 10, color = "blue"), show.legend = FALSE, arrow = a) +
    geom_node_point(color = "lightblue", size = 16) +
    geom_node_text(aes(label = name, size = 20), repel = TRUE) +
    theme_void() +
    theme(legend.position="none")
}
obama_tweets_df  %>% count_bigrams()  %>% head(50) %>% visualize_bigrams()

Word Frequencies

Now we’ll count word frequencies and create a word cloud.

tweets <- obama_tweets_df

reg <- "([^A-Za-z\\d#@']|'(?![A-Za-z\\d#@]))"
tweet_words <- tweets %>%
  filter(!str_detect(text, '^"')) %>%
  mutate(text = str_replace_all(text, "https://t.co/[A-Za-z\\d]+|&amp;", "")) %>%
  unnest_tokens(word, text, token = "regex", pattern = reg) %>%
  filter(!word %in% stop_words$word,
         str_detect(word, "[a-z]"))

tweet_words <- tweet_words %>% group_by(word) %>% tally() %>% mutate(freq = n) %>% select(word, freq) %>% arrange(desc(freq))

tweet_words %>% head(20) %>% kable(row.names=TRUE)
word freq
1 senate 68
2 #doyourjob 61
3 leaders 57
4 president 50
5 obama 48
6 court 43
7 supreme 42
8 #actonclimate 40
9 climate 39
10 change 35
11 judge 33
12 garland 27
13 hearing 25
14 add 21
15 economy 20
16 progress 20
17 agree 19
18 join 19
19 fair 18
20 @ofa 18 
wordcloud2(tweet_words, size = 0.6)


Sentiment

Now we’ll analyze sentiment.

obama_sentiment <- tweet_words %>%
  inner_join(get_sentiments("bing"))

obama_sentiment %>%
  group_by(sentiment) %>%
  top_n(10, freq) %>%
  arrange(freq) %>%
  ungroup() %>%
  mutate(word = reorder(word, freq)) %>%
  ggplot(aes(word, freq, fill = sentiment)) +
  geom_col(show.legend = FALSE) +
  facet_wrap(~sentiment, scales = "free_y") +
  labs(y = "Contribution to sentiment",
       x = NULL) +
  coord_flip()