remove extraneous images

get twitter data.R

@@ -0,0 +1,160 @@
+library(tidyverse)
+library(tidytext)
+library(rvest)
+library(drlib)
+library(rtweet)
+library(httpuv)
+# install.packages("devtools")
+# library(devtools)
+# install_github("dgrtwo/drlib")
+library(drlib)
+library(widyr)
+
+source(tokens.R)
+
+# list of prominent Dallas media members
+# from https://twitter.com/advocamentum/lists/news-media/members
+advocamentum_news_media <- lists_members(owner_user = "Advocamentum", slug = "news-media")
+
+# add Wylie to advocamentum's list
+advocamentum_news_media <- advocamentum_news_media %>%
+  add_row(name = "Wylie H. Dallas",
+          screen_name = "Wylie_H_Dallas") %>%
+  add_row(name = "Philip Kingston",
+          screen_name = "PhilipTKingston")
+
+# Download ~3200 from each account
+tweets <- map_df(advocamentum_news_media$screen_name,
+                get_timeline,
+                n = 3200)
+
+wylie_tweets <- get_timeline("Wylie_H_Dallas", n = 3200)
+
+save(tweets, file="tweets_20190507.Rda")
+save(wylie_tweets, file="wylie_tweets_20190507.Rda")
+
+kingston_tweets <- get_timeline("PhilipTKingston", n = 3200)
+tweets <- tweets %>%
+  bind_rows(kingston_tweets)
+
+load(file = "tweets.Rda")
+load(file = "wylie_tweets.Rda")
+
+# merge data frames together
+tweets <- bind_rows(tweets, wylie_tweets)
+
+# create plot showing time of day when everyone tweets
+
+# first get the hour
+tweets$hour <- as.numeric(strftime(tweets$created_at, format = "%H")) +
+  as.numeric(strftime(tweets$created_at, format = "%M")) / 60
+
+tweets %>%
+  filter(screen_name %in% c("JimSchutze",
+                            "Wylie_H_Dallas",
+                            "RobertWilonsky")) %>%
+  ggplot(aes(x = hour)) +
+  geom_histogram(bins = 24 * 6) +
+  facet_grid(~screen_name)
+
+# let's do the above on everyone individually
+for (i in unique(tweets$screen_name)) {
+  tweets %>%
+    filter(screen_name == i |
+             screen_name == "Wylie_H_Dallas") %>%
+    ggplot(data = .[screen_name == i,], aes(x = hour)) +
+    geom_histogram(bins = 24 * 6) +
+    xlim(0, 24) +
+    xlab("hour of day") +
+    ylab("count of tweets") +
+    ggtitle(paste(advocamentum_news_media[advocamentum_news_media$screen_name == i,]$name, "'s tweet time histogram", sep = ""))
+  ggsave(paste(i, ".png"))
+}
+
+i = "PhilipTKingston"
+tweets %>%
+  filter(screen_name == i) %>%
+  ggplot(aes(x = hour)) +
+  geom_histogram(bins = 24 * 6) +
+  xlim(0, 24) +
+  xlab("hour of day") +
+  ylab("count of tweets") +
+  ggtitle(paste(advocamentum_news_media[advocamentum_news_media$screen_name == i,]$name, "'s tweet time histogram", sep = ""))
+
+
+tweets %>%
+  filter(screen_name == i |
+           screen_name == "Wylie_H_Dallas") %>%
+  ggplot(data = .[.$screen_name == i,], aes(x = hour)) +
+  geom_histogram(bins = 24 * 6) +
+  xlim(0, 24) +
+  xlab("hour of day") +
+  ylab("count of tweets") +
+  ggtitle(paste(advocamentum_news_media[advocamentum_news_media$screen_name == i,]$name, "'s tweet time histogram", sep = ""))
+ggsave(paste(i, ".png"))
+
+# do word analysis
+reg <- "([^A-Za-z\\d#@']|'(?![A-Za-z\\d#@]))"
+
+tweet_words <- tweets %>%
+  filter(!is_retweet) %>%
+  arrange(created_at) %>%
+  distinct(text, .keep_all = TRUE) %>%
+  select(screen_name, status_id, text) %>%
+  mutate(text = str_replace_all(text, "https?://t.co/[A-Za-z\\d]+|&amp;", "")) %>%
+  unnest_tokens(word, text, token = "regex", pattern = reg) %>%
+  filter(str_detect(word, "[a-z]")) %>%
+  filter(!word %in% stop_words$word)
+
+tweet_words %>%
+  filter(!word %in% stop_words$word) %>%
+  count(word, sort = TRUE) %>%
+  head(16) %>%
+  mutate(word = reorder(word, n)) %>%
+  ggplot(aes(word, n)) +
+  geom_col() +
+  coord_flip() +
+  labs(y = "# of uses among staff Twitter accounts")
+
+word_counts <- tweet_words %>%
+  count(screen_name, word, sort = TRUE)
+
+# Compute TF-IDF using "word" as term and "screen_name" as document
+word_tf_idf <- word_counts %>%
+  bind_tf_idf(word, screen_name, n) %>%
+  arrange(desc(tf_idf))
+
+similarity <- word_tf_idf %>%
+  pairwise_similarity(screen_name, word, tf_idf, upper = FALSE, sort = TRUE) %>%
+  filter(item1 == "Wylie_H_Dallas" |
+           item2 == "Wylie_H_Dallas")
+
+# get technology data
+tweets_datefiltered %>%
+  group_by(screen_name, source) %>%
+  summarize(count = n()) %>%
+  ungroup() %>%
+  pairwise_similarity(screen_name, source, count, upper = FALSE, sort = TRUE) %>%
+  filter(item1 == "Wylie_H_Dallas" |
+           item2 == "Wylie_H_Dallas")
+
+# pairwise similarity by hour
+hour_matches <- tweets_datefiltered %>%
+  mutate(hour_number = as.numeric(strftime(tweets_datefiltered$created_at, format = "%H"))) %>%
+  group_by(screen_name, hour_number) %>%
+  summarize(count = n()) %>%
+  ungroup() %>%
+  pairwise_similarity(screen_name, hour_number, count, upper = FALSE, sort = TRUE) %>%
+  filter(item1 == "Wylie_H_Dallas" |
+         item2 == "Wylie_H_Dallas")
+
+# pairwise similarity by date
+date_matches <- tweets_datefiltered %>%
+  mutate(date_string = as.Date(tweets_datefiltered$created_at)) %>%
+  group_by(screen_name, date_string) %>%
+  summarize(count = n()) %>%
+  ungroup() %>%
+  pairwise_similarity(screen_name, date_string, count, upper = FALSE, sort = TRUE) %>%
+  filter(item1 == "Wylie_H_Dallas" |
+           item2 == "Wylie_H_Dallas")
+