Twitter Data-Mini Project

Transcript

1. Data Mini-Project Nitin Jain Pavan Kumar Geoffrey Jacoby Abhishek Nandakumar
3. 151.78240 35.45195 68.79268 79.56177
4. p-value ~ 1.7e−18
5. ordinary people hollywood stars
6. “I want”
7. San Francisco Memphis Things for Christmas Something (describe it) Someone (to do something) Ice Cream Love Things for Christmas Money Pizza Someone (to do something..) (Some kind of) House
8. words common to SF + Memphis p-value = 0.127 Difference insignificant.
9. all unique words p-value = 3.17e−07 Difference meaningful.
10. Potential confounds Wants for Christmas. ! Words occurring once. ! Context changes meaning: ! I seriously think I want to hire someone to brush my hair for me after I get out of the shower. DAMN. So many tangles. #lazy ! I want someone to fall in love with the way I laugh and fall in love with my smile.
11. Follower Count Following Count Listed (count) Tweet count URLs in 100 most recent tweets Gender 118 celebrities
13. Outliers
14. Barack Obama Arnold Schwarzenegger Al Gore Senator John Mccain Sarah Palin Hillary Clinton John Boehner Joe Biden Politicians 40,185,454 2,925,511 2,701,171 1,849,471 998,646 945,905 585,267 519,641
15. Tim O’Reilly Deepak Chopra Guy Kawasaki John C. Maxwell Beth Kanter Authors 1,736,824 1,694,682 1,395,960 869,970 403,402
16. Kaka Shaquille O' Neal Lance Armstrong Rubens Barrichello Mark Sanchez Sports 17,452,018 7,843,305 3,918,367 2,020,439 832,289
19. 118 → 68 celebrities Linear model found variables significant.
20. Adjusted R2: 0.5444 ! p-values ! Number of Following 0.00892 Number of Tweets 0.02422 Public List Appearance 2.39e−10)
21. Margin of error above 10% ! Sampling bias ! Missed features Time since on Twitter Rise in popularity because of recent event Issues