Assignment-4 MBA 676
YUSUF SULTAN
0.1 Introductions
“IF DONALD TRUMP loves one thing in this world, it’s Donald Trump. If he loves two things, a close second-place goes to his Twitter feed. Unfortunately, in addition to Trump’s loyal armies of patriots and trolls, Twitter also hosts both liberals and members of the lying liberal media—people who do not show President Trump the respect he deserves. And for that, they must be blocked.” wired, well everyone has his opinion.Since almost all the media mentions the relationship between president Trump and twitter and our assignment based on using twitter. I picked the two # words trump and Obama and make Twitter Data Analysis compare those words.In this assignment, we will make campaigning about the current tweets on two words trump and Obama.
0.2 Access Twitter Data
setup_twitter_oauth(consumer_key, consumer_secret, access_token, access_secret)## [1] "Using direct authentication"0.3 Search Twitter
The Twitter API returns the tweets as a json stream and makes search on Twitter.In this function, we are going to make search in Twitter on two words or hashtags trump and Obama in those 24 hours or current tweets
#hashtage trump
trump_tweets <- 1000
trump <- searchTwitter('#trump', n = trump_tweets)
head(trump)## [[1]]
## [1] "ElderJade: RT @TalbertSwan: @COGICFamily @realDonaldTrump Its only a big deal with #Trump when a Democrat gets accused of it! He dismisses the 16 wom
"
##
## [[2]]
## [1] "TimmieTadpole: RT @Emolclause: #BREAKING:.@GOP is actively HELPING #Trump PROFIT from his Presidency. This is the UNFETTERED #CORRUPTION we're accustomed
"
##
## [[3]]
## [1] "Anita67789399: RT @kwilli1046: \"I cannot get enough of Donald #Trump's Twitter feed. If you are not enjoying it, I feel bad. This is the best thing ever
"
##
## [[4]]
## [1] "SGuilbaud22: RT @amjoyshow: Women who have accused #Trump of sexual misconduct. #AMJoy https://t.co/MSTf6bjETG"
##
## [[5]]
## [1] "cdeltess: RT @TalbertSwan: @COGICFamily @realDonaldTrump Its only a big deal with #Trump when a Democrat gets accused of it! He dismisses the 16 wom
"
##
## [[6]]
## [1] "bgood12345: RT @Trumpfan1995: How will you mostly spend your Sunday? Do vote and retweet. #MAGA #Trump #NFL"#hashtag obama
obama_tweets <- 1000
obama <- searchTwitter('#obama', n = obama_tweets)
head(obama)## [[1]]
## [1] "RealJohhnieDoe: <ed><U+00A0><U+00BD><ed><U+00BA><U+00A8>MUST WATCH! The 7th Floor CIA officer that developed the CIA Anti-Terror program speaks out on:\n#Obama \n#Clinton
https://t.co/XhjQR2FnsK"
##
## [[2]]
## [1] "Earthboy4life: RT @KNP2BP: #Obama left #Tahmooressi in Mexico & #Wambier in NK\n\nMarine released b/c of @greta , Gov Richardson NM & Reps. Royce (R-CA) Sal
"
##
## [[3]]
## [1] "jemz1113: RT @IncognitoPatrio: HERE'S THE REAL #RussianCollusion! #Obama s Russian Collusion is ON VIDEO https://t.co/EyW8sfuSyh via @truthfeednews"
##
## [[4]]
## [1] "kakeenan: RT @wavetossed: #qanon #Obama Barry Soetoro used a #Kenya story to hide the fact that he was #born in #Indonesia and a #Columbia #Universit
"
##
## [[5]]
## [1] "Kayem623: RT @KNP2BP: #Obama left #Tahmooressi in Mexico & #Wambier in NK\n\nMarine released b/c of @greta , Gov Richardson NM & Reps. Royce (R-CA) Sal
"
##
## [[6]]
## [1] "PatriotSteve4U: RT @KNP2BP: #Obama left #Tahmooressi in Mexico & #Wambier in NK\n\nMarine released b/c of @greta , Gov Richardson NM & Reps. Royce (R-CA) Sal
"0.4 Convert the list
Convert the list to a data frame using twListToDF().
trump_df <- twListToDF(trump)
obama_df <- twListToDF(obama)0.5 Data Frame Contains
Ever HTML page has a meta tag as describe the page and usually it is the website link of the page names title numbers author The data frame contains the tweet along with 15 other metadata items including the username, the retweet count, the platform/application the tweet was created on
Trump
trump_df %>% group_by(statusSource) %>%
summarize(n = n()) %>%
arrange(desc(n)) %>%
top_n(10)## Selecting by n## # A tibble: 10 x 2
## statusSource
## <chr>
## 1 "<a href=\"http://twitter.com/download/iphone\" rel=\"nofollow\">Twitter fo
## 2 "<a href=\"http://twitter.com/download/android\" rel=\"nofollow\">Twitter f
## 3 "<a href=\"http://twitter.com\" rel=\"nofollow\">Twitter Web Client</a>"
## 4 "<a href=\"http://twitter.com/#!/download/ipad\" rel=\"nofollow\">Twitter f
## 5 "<a href=\"https://mobile.twitter.com\" rel=\"nofollow\">Twitter Lite</a>"
## 6 "<a href=\"https://about.twitter.com/products/tweetdeck\" rel=\"nofollow\">
## 7 "<a href=\"https://ifttt.com\" rel=\"nofollow\">IFTTT</a>"
## 8 "<a href=\"https://github.com/dekuplant?tab=repositories\" rel=\"nofollow\"
## 9 "<a href=\"http://grrrgrumbles.me/\" rel=\"nofollow\">Robocuck</a>"
## 10 "<a href=\"http://www.labnol.org\" rel=\"nofollow\">Jess Ella</a>"
## # ... with 1 more variables: n <int>Obama
obama_df %>% group_by(statusSource) %>%
summarize(n = n()) %>%
arrange(desc(n)) %>%
top_n(10)## Selecting by n## # A tibble: 13 x 2
## statusSource
## <chr>
## 1 "<a href=\"http://twitter.com\" rel=\"nofollow\">Twitter Web Client</a>"
## 2 "<a href=\"http://twitter.com/download/iphone\" rel=\"nofollow\">Twitter fo
## 3 "<a href=\"http://twitter.com/download/android\" rel=\"nofollow\">Twitter f
## 4 "<a href=\"http://twitter.com/#!/download/ipad\" rel=\"nofollow\">Twitter f
## 5 "<a href=\"https://mobile.twitter.com\" rel=\"nofollow\">Twitter Lite</a>"
## 6 "<a href=\"https://ifttt.com\" rel=\"nofollow\">IFTTT</a>"
## 7 "<a href=\"https://about.twitter.com/products/tweetdeck\" rel=\"nofollow\">
## 8 "<a href=\"http://www.hootsuite.com\" rel=\"nofollow\">Hootsuite</a>"
## 9 "<a href=\"http://bufferapp.com\" rel=\"nofollow\">Buffer</a>"
## 10 "<a href=\"http://ingminds.com\" rel=\"nofollow\">Zlando</a>"
## 11 "<a href=\"http://leadstories.com\" rel=\"nofollow\">Lead Stories Feed Publ
## 12 "<a href=\"http://www.brianbrown.net/\" rel=\"nofollow\">BBN - SNAP</a>"
## 13 "<a href=\"http://www.twitter.com\" rel=\"nofollow\">Twitter for Windows</a
## # ... with 1 more variables: n <int>totalTrump <- trump_df %>% summarise(total = sum(n()))
totalObama <- obama_df %>% summarise(total = sum(n()))
trump_df$created <- as.POSIXct(trump_df$created, format="%Y-%m-%d %H%M%S")
trump_df_hour <- trump_df %>% mutate(formatD = format(round(trump_df$created, units="hours"), format="%H:%M")) %>% group_by(formatD) %>% summarize(n = n(), pct = n/2368*100)
obama_df$created <- as.POSIXct(obama_df$created, format="%Y-%m-%d %H%M%S")
obama_df_hour <- obama_df %>% mutate(formatD = format(round(obama_df$created, units="hours"), format="%H:%M")) %>% group_by(formatD) %>% summarize(n = n(), pct = n/10)
trump_df_hour$U.S.President <- "Donald Trump"
obama_df_hour$U.S.President <- "Barack Obama"
Total_hour <- rbind(trump_df_hour, obama_df_hour)0.6 Plotting the data
0.6.1 By Platforms
compare platforms of trump hashtag and obama took from the current twetts during the day .
ggplot(Total_hour, aes(x = formatD , y = pct, fill = U.S.President)) +
geom_bar(stat = "identity", position = "dodge") +
xlab("Today") +
ylab("% tweets") +
theme(axis.text.x = element_text(angle = 90, hjust = 1))
nrc <- sentiments %>%
filter(lexicon == "nrc") %>%
select(word, sentiment)0.6.2 Clean Data
process of detecting and removing anything is not word by using the tidytext library
reg <- "([^A-Za-z\\d#@']|'(?![A-Za-z\\d#@]))"
trump_words <- trump_df %>%
filter(!str_detect(text, '^"')) %>%
mutate(text = str_replace_all(text, "https://t.co/[A-Za-z\\d]+|&", "")
) %>% unnest_tokens(word, text, token = "regex", pattern = reg) %>%
filter(!word %in% stop_words$word, str_detect(word, "[a-z]"))
obama_words <- obama_df %>%
filter(!str_detect(text, '^"')) %>%
mutate(text = str_replace_all(text, "https://t.co/[A-Za-z\\d]+|&", "")
) %>% unnest_tokens(word, text, token = "regex", pattern = reg) %>%
filter(!word %in% stop_words$word, str_detect(word, "[a-z]"))
#Join nrc to words dataframes for Science and Art
trump_sentiments <- trump_words %>% inner_join(nrc, by = "word")
obama_sentiments <- obama_words %>% inner_join(nrc, by = "word")
trump_sentiments %>% group_by(sentiment) %>% summarize(n = n()) %>% arrange(desc(n))## # A tibble: 10 x 2
## sentiment n
## <chr> <int>
## 1 negative 636
## 2 positive 593
## 3 trust 471
## 4 anger 431
## 5 fear 425
## 6 anticipation 356
## 7 joy 327
## 8 sadness 294
## 9 surprise 267
## 10 disgust 229obama_sentiments %>% group_by(sentiment) %>% summarize(n = n()) %>% arrange(desc(n))## # A tibble: 10 x 2
## sentiment n
## <chr> <int>
## 1 negative 578
## 2 trust 436
## 3 fear 412
## 4 surprise 367
## 5 positive 363
## 6 anticipation 341
## 7 sadness 200
## 8 anger 193
## 9 disgust 193
## 10 joy 1350.6.3 By Sentiments
trump_sentiments$U.S.President <- "Donald Trump"
obama_sentiments$U.S.President <- "Barack Obama"
Sentiments <- rbind(trump_sentiments, obama_sentiments)
Sentiments_df <- Sentiments %>%
group_by(U.S.President, sentiment) %>%
summarize(n = n()) %>%
mutate(pct = n/sum(n))ggplot(Sentiments_df, aes(x = sentiment, y = pct, fill = U.S.President)) +
geom_bar(stat = "identity", position = "dodge") +
xlab("Sentiment") +
ylab("% Tweets") +
theme(axis.text.x = element_text(angle = 90, hjust = 1))
0.7 Wordclouds
0.7.1 Trump Words
trump_wc <- trump_sentiments %>% group_by(word) %>% summarize(count = n()) %>% arrange(desc(count))
wordcloud(trump_wc$word, trump_wc$count, random.order=FALSE, scale=c(5, .8), use.r.layout=TRUE,colors = brewer.pal(6, "Dark2"), max.words=90)
0.7.2 Obama Words
obama_wc <- obama_sentiments %>% group_by(word) %>% summarize(count = n()) %>% arrange(desc(count))
wordcloud(obama_wc$word, trump_wc$count, random.order=FALSE, scale=c(5, .8), use.r.layout=TRUE,colors = brewer.pal(6, "Dark2"), max.words=90)
0.8 Sentiment Score
0.8.1 Time Day
ggplot(data=Total_hour,aes(x =formatD,y = n))+
geom_bar(aes(fill=formatD),stat="identity")+
theme(legend.position = "none")+
xlab("sentiment")+ylab("Score")+ggtitle("Total Sentiment Score During the Day")
0.8.2 Words Score
ggplot(data=Sentiments_df,aes(x =sentiment,y = n))+
geom_bar(aes(fill=sentiment),stat="identity")+
theme(legend.position = "none")+
xlab("sentiment")+ylab("Score")+ggtitle("Total Sentiment Score Based on words Tweets")