Project Description
This [tweet](https://twitter.com/tvaziri/status/762005541388378112) containing a hypothesis about Donald Trump's Twitter account needs to be investigated with data:
Others have [explored Trump’s timeline](http://www.cnet.com/news/trumps-tweets-android-for-nasty-iphone-for-nice/) and noticed this tends to hold up. And Trump himself [did indeed tweet from a Samsung Galaxy](http://www.theverge.com/2015/10/5/9453935/donald-trump-twitter-strategy) until [March 2017](https://www.recode.net/2017/5/27/15705090/president-donald-trump-twitter-android-iphone-ios-samsung-galaxy-security-hacking). But how could it be examined quantitatively? In this project, you will apply text mining and sentiment analysis to determine whether or not Trump does indeed write the angrier, Android tweets The dataset used in this project is from [The Trump Twitter Archive](http://www.trumptwitterarchive.com/) by Brendan Brown, which contains all 35,000+ tweets from the [@realDonaldTrump](https://twitter.com/realDonaldTrump/) Twitter account from 2009 (the year Trump sent his [first tweet](https://www.businessinsider.de/donald-trump-first-tweet-2017-5)) through 2018.Every non-hyperbolic tweet is from iPhone (his staff).
— Todd Vaziri (@tvaziri) August 6, 2016
Every hyperbolic tweet is from Android (from him).
Project Tasks
- 1The tweets
- 2Clean those tweets
- 3Is "time" the giveaway?
- 4The quote tweet is dead
- 5Links and pictures
- 6Comparison of words
- 7Most common words
- 8Common words: Android vs. iPhone (i)
- 9Common words: Android vs. iPhone (ii)
- 10Adding sentiments
- 11Android vs. iPhone sentiments
- 12Conclusion: The ghost in the political machine
Technologies
R
David Robinson
Principal Data Scientist at Heap
Dave is the Principal Data Scientist at Heap. He has worked as a data scientist at DataCamp and Stack Overflow, and received his PhD in Quantitative and Computational Biology from Princeton University. Follow him at @drob on Twitter or on his blog, Variance Explained.