3_6_rtweet.Rmd

---
title: "3_6_rtweet"
output: 
  html_notebook: 
    code_folding: hide
    number_sections: true
    fig_caption: yes
    highlight: textmate
    theme: simplex
    toc: yes
editor_options: 
  chunk_output_type: inline
---

# INFO
* http://rtweet.info/

## Create twitter app to get data 

Follow the instruction http://rtweet.info/articles/auth.html

1. Visit http://apps.twitter.com
2. Create New App
3. Fill in app Name, Description, Website (such as http://www.ntu.edu.tw), callback URL(for RStudio, should be http://127.0.0.1:1410), **SERIOUSLY**!
4. View your app and select the tab "Keys and Access Tokens"
5. Copy your *consumer Key* and *Consumer Secret*. 


```{r message=FALSE, warning=FALSE, paged.print=FALSE}
# install.packages("rtweet")
library(rtweet)
library(stringr)
library(tidyverse)
library(tidytext)
browseURL("http://rtweet.info/")
```

# Setting tokens
* http://rtweet.info/articles/auth.html
```{r}
browseURL("http://rtweet.info/articles/auth.html")
```


# Search tweets
```{r}
tweets <- search_tweets(
  "#taiwan", n = 10000, include_rts = FALSE
)
tweets %>% names
```

# Filtering data
```{r}
filtered.df <- tweets %>%
    filter(!str_detect(text, '^"')) %>%
	mutate(timestamp = as.POSIXct(created_at)) %>%
	select("status_id", "user_id", "screen_name", "text", 
		   "source", "reply_to_user_id", "reply_to_screen_name",
		   "is_quote", "is_retweet")
head(filtered.df, n=3)
```

# Get more hashtags from text
```{r}
unnested.df <- filtered.df %>%
    unnest_tokens(word, text, 
                  token = "regex", pattern = "[^A-Za-z\\d#@']", 
                  drop=FALSE) 

hashtags.df <- unnested.df %>%
	filter(str_detect(word, "^#")) %>%
	filter(nchar(word) > 1) %>%
	mutate(text = str_to_lower(text)) %>%
	filter(str_detect(text, "#dominicanrepublic"))
```


# Descriptive analysis of hashtags
```{r}
hashtags.df %>% count(word, sort = T)

top_user <- hashtags.df %>% count(user_id, sort = T) %>%
	slice(1:20)

users <- lookup_users(top_user$user_id, parse = T)

```

# Assignment: tweets data
* Propose and run your idea to analyze tweets
* What kinds of topic do you watch?


# Get data by specific token
```{r}
token1 <- create_token(app = "YOUR_APP_NAME",
             consumer_key = "YOUR_API_KEY",
             consumer_secret = "YOUR_API_SECRET")
fls <- get_followers(x, n = 75000, token = token1)
```