Graph top words from a search query

twitterfunctions.Rmd

```{r}
library(rtweet)
library(twitteR)
library(tidyverse)
library(tidytext)
library(httpuv)
data(stop_words)
```

```{r}
#insert your own tokens
appname <- "xxxxx"
key <- "xxxxx"
secret <- "xxxxxx"

twitter_token <- create_token(
  app = appname,
  consumer_key = key,
  consumer_secret = secret)
```

```{r}
#top words scraper
#top_tweets[1] is the search term, include quotes
#top_tweets[2] is number of results
#top_tweets[3] is number of recent tweets, 18,000 
#top_tweets[4] is basic colour scheme (default black)
#add any junk terms to the word list in custom_stop_words

top_words <- function(keyword, topn, total, visual = "black"){

  tweet_data <- search_tweets(keyword, n = total, include_rts = F, type = "recent", retryonratelimit = T)

  custom_stop_words <- bind_rows(data_frame(word = c(gsub("#", "", keyword),"t.co", "https", "rt", "amp"), 
                                          lexicon = c("custom")), stop_words)

  my_title <- paste("Online Discussion about", keyword)
  my_subtitle <- paste("based on", total, "recent tweets")

tweet_data %>%
  unnest_tokens(word, text) %>%
  anti_join(custom_stop_words) %>%
  count(word, sort = T) %>%
  head(topn) %>%
ggplot(aes(x = reorder(word, n), y = n, fill = "blue")) + 
  geom_bar(show.legend = F, stat = "identity", width = 0.8) +
  scale_fill_manual(values = visual) + 
  labs(title = my_title, subtitle = my_subtitle, y = "Number of Mentions", x = NULL) +
  coord_flip() +
  theme_classic() +
  theme(plot.title=element_text(family='', face='bold', colour = visual, size=16))
}
```

```{r}
#example
top_words("#onpoli", 25, 1000, "red")
```

```{r}
#timeline scraper

timeline_tweets <- function(keyword, topn, total, visual = "black"){
  tweet_data <- get_timeline(keyword, n = total)

  custom_stop_words <- bind_rows(data_frame(word = c(gsub("#", "", keyword),"t.co", "https", "rt", "amp"), 
                                          lexicon = c("custom")), stop_words)

  my_title <- paste0("Most frequent word usage by @", keyword)
  my_subtitle <- paste("based on", total, "recent tweets")

tweet_data %>%
  unnest_tokens(word, text) %>%
  anti_join(custom_stop_words) %>%
  count(word, sort = T) %>%
  head(topn) %>%
ggplot(aes(x = reorder(word, n), y = n, fill = "blue")) + 
  geom_bar(show.legend = F, stat = "identity", width = 0.8) +
  scale_fill_manual(values = visual) + 
  labs(title = my_title, subtitle = my_subtitle, y = "Number of Mentions", x = NULL) +
  coord_flip() +
  theme_classic() +
  theme(plot.title=element_text(family='', face='bold', colour = visual, size=16))
}
```

```{r}
timeline_tweets("Kathleen_Wynne", 15, 500, "red")
timeline_tweets("AndreaHorwath", 15, 500, "orange")
timeline_tweets("fordnation", 15, 500, "blue")
```

```{r}
#keywords of conversation
wynne <- "@kathleen_wynne OR kathleen wynne OR wynne OR @OntLiberal"
ford <- "@fordnation OR doug ford OR @OntarioPCParty)"
horvath <- "@andreaHorwath OR andrea horwath OR horwath OR @OntarioNDP"


conversation <- function(keyword, topn, total, visual = "black"){

  tweet_data <- search_tweets(keyword, n = total, type = "recent", include_rts = F, retryonratelimit = TRUE)

  custom_stop_words <- bind_rows(data_frame(word = c(gsub("#", "", keyword), "t.co", "https", "rt", "amp"), 
                                          lexicon = c("custom")), stop_words)

  my_title <- paste("Online Discussion Surrounding", keyword)
  my_subtitle <- paste("based on", total, "recent tweets")

tweet_data %>%
  unnest_tokens(word, text) %>%
  anti_join(custom_stop_words) %>%
  count(word, sort = T) %>%
  head(topn) %>%
ggplot(aes(x = reorder(word, n), y = n, fill = "blue")) + 
  geom_bar(show.legend = F, stat = "identity", width = 0.8) +
  scale_fill_manual(values = visual) + 
  labs(title = my_title, subtitle = my_subtitle, y = "Number of Mentions", x = NULL) +
  coord_flip() +
  theme_classic() +
  theme(plot.title=element_text(family='', face='bold', colour = visual, size=16))
}

```

```{r}
conversation(wynne, 25, 1000, "red")
conversation(ford, 25, 1000, "blue")
conversation(horvath, 25, 1000, "orange")
```

```{r}
#bigrams of conversation
wynne <- "@kathleen_wynne OR kathleen wynne OR wynne OR @OntLiberal"
ford <- "@fordnation OR doug ford OR @OntarioPCParty)"
horvath <- "@andreaHorwath OR andrea horwath OR horwath OR @OntarioNDP"

conversation2 <- function(keyword, total, visual = "black"){
  
tweet_data <- search_tweets(keyword, n = total, type = "recent", include_rts = F, retryonratelimit = TRUE)

custom_stop_words <- bind_rows(data_frame(bigram = c("https t.co", "of the", "for the", "and the", "to be", "to the", "in the", "is a", "is the"), 
                                          lexicon = c("custom")), stop_words)

  my_title <- paste("Bigrams on Discussion Surrounding", keyword)
  my_subtitle <- paste("based on", total, "recent tweets")

tweet_data %>%
  unnest_tokens(bigram, text, token = "ngrams", n = 2, collapse = F) %>%
  anti_join(custom_stop_words) %>%
  count(bigram, sort = T) %>%
  head(15) %>%
ggplot(aes(x = reorder(bigram, n), y = n, fill = "blue")) + 
  geom_bar(show.legend = F, stat = "identity", width = 0.8) +
  scale_fill_manual(values = visual) + 
  labs(title = my_title, subtitle = my_subtitle, y = "Number of Mentions", x = NULL) +
  coord_flip() +
  theme_classic() +
  theme(plot.title=element_text(family='', face='bold', colour = visual, size=16))
}

```

```{r}
conversation2(wynne, 1000, "red")
conversation2(ford, 1000, "blue")
conversation2(horvath, 1000, "orange")
```

Updated ggplot visuals and cleaned some consistencies with alignment.
Planning to remove any twitter handles from conversation by str_replace anything with @