R Notebook

I chose to do my search on Jimmy Fallon. This first chunk retrieves all of the data on tweets that others are making about Jimmy fallon.

fallon <- search_tweets("Jimmy Fallon", n = 1000, include_rts = F)

Downloading [=======>---------------------------------]  20%
Downloading [===========>-----------------------------]  30%
Downloading [===============>-------------------------]  40%
Downloading [===================>---------------------]  50%
Downloading [========================>----------------]  60%
Downloading [============================>------------]  70%
Downloading [================================>--------]  80%
Downloading [====================================>----]  90%
Downloading [=========================================] 100%

Using the above data, this next chunk creates a table of the top most popular tweets with “Jimmy Fallon”.

fallon %>% 
  select(text, retweet_count) %>%    
  top_n(25) %>%                       # get the top 25 most popular tweets
  arrange(-retweet_count) %>%         # sort in descending order of popularity
  datatable()

Selecting by retweet_count

This next chunk then retrieves all the tweets that jimmy fallon himself has made on his own twitter page.

fallon_tweets <- get_timeline("jimmyfallon", n = 5000)

Using the information retrieved from above, this chunk finds the hashtags Jimmy Fallon uses the most and creates a table.

fallon_tweets %>% 
  select(hashtags) %>%                   # Focus on the hashtags
  unnest() %>%                           # Separate multiple hashtags
  mutate(hashtags = tolower(hashtags)) %>%      # make all hashtags lowercase
  count(hashtags, sort=TRUE) %>%                # count how often they appear
  datatable()                                   # create an interactive table

`cols` is now required when using unnest().
Please use `cols = c(hashtags)`

This next chunk finds how many tweets per day and organizes it into a table.

fallon_tweets %>% 
  group_by(day = date(created_at)) %>%    # extract the date, group by it
  summarize(tweets_per_day = n())         # count the number of tweets each day

This chunk now finds the average number of tweets Jimmy Fallon makes per day.

fallon_tweets %>% 
  group_by(day = date(created_at)) %>%    # extract the date, group by it
  summarize(tweets_per_day = n()) %>%        # count the number of tweets each day
  summarize(mean(tweets_per_day))

Now, using plotly this next chunk shows the number of tweets per day from Jimmy Fallon in a histogram rather than a table.

fallon_tweets %>%
  mutate(day = date(created_at)) %>% 
  plot_ly(x = ~day) %>%                                        
  add_histogram() %>%
  layout(title = "Number of tweets from @jimmyfallon")

This next chunk then uses the data to find how many tweets per hour of the day Jimmy Fallon tweets. It also takes into account his time zone which I believe is Los Angeles.

fallon_tweets %>% 
  mutate(time = with_tz(created_at, "America/Los_Angeles")) %>% 
  mutate(time = hour(time)) %>% 
  count(time) %>% 
  datatable(options = (list(pageLength = 24)), rownames = F)

This next chunk then changes the table into a plotly histogram of how many tweets Jimmy fallon makes at each hour.

fallon_tweets %>% 
  mutate(time = with_tz(created_at, "America/Los_Angeles")) %>%    # convert to Los Angeles time zone
  mutate(time = hour(time)) %>%                                 # extract the hour
  plot_ly(x = ~time) %>%                                        # create plotly graph
  add_histogram() %>%                                          # make histogram
  layout(title = "When Does @jimmyfallon Tweet?", 
         xaxis = list(title = "Time of Day (0 = midnight)"),
         yaxis = list(title = "Number of Tweets"))

NA

This next chunk then organizes the tweets by day, showing on average how many tweets Jimmy Fallon makes on each day of the week.

fallon_tweets %>% 
  mutate(Day = wday(created_at,           # find the weekday that the tweet was created
                    label = T)) %>%       # use labels (Sun, Mon, etc) rather than numbers
  count(Day) %>%                          # count the number of tweets each day
  datatable(rownames = F)

The next chunk then converts the table into a histogram showing the same data of tweets each weekday.

fallon_tweets %>% 
  mutate(Day = wday(created_at,           # find the weekday that the tweet was created
                    label = T)) %>%       # use labels (Sun, Mon, etc) rather than numbers
   plot_ly(x = ~Day) %>%                                        # create plotly graph
  add_histogram() %>%                                          # make histogram
  layout(title = "When Does @jimmyfallon Tweet?", 
         xaxis = list(title = "Days of the week"),
         yaxis = list(title = "Number of Tweets"))

This last chunk combines the information from the histograms and creates a heat map showing which hour of each day Jimmy fallon tweets the most.

fallon_tweets %>% 
  mutate(day = wday(created_at, label = T)) %>% 
  mutate(hour = hour(with_tz(created_at, "America/Los_Angeles"))) %>% 
  plot_ly(x = ~day, y = ~hour) %>% 
  add_histogram2d(nbinsx = 7, nbinsy = 24) %>%
  layout(title = "When Does @jimmyfallon Tweet?", 
         xaxis = list(title = "Days of the week"),
         yaxis = list(title = "hour"))

NA