NeutralOpinion 
Visualizing how people really feel about net neutrality. 
! 
! 
! 
! 
Geneviève Smith | Insight Data Science NYC | August 2014
Working demo
Working demo
Working demo
Working demo
The FCC doesn’t usually get this massive a response 
Number of comments/ 
complaints
The FCC doesn’t usually get this massive a response 
Number of comments/ 
complaints
neutralopinion.com
Raw comments (~250,000 out of 1 million)
Raw comments (~250,000 out of 1 million) 
Census tallies of houses with internet access
Raw comments (~250,000 out of 1 million) 
Census tallies of houses with internet access 
NLTK 
(natural language processing toolkit) 
to build term-document matrix 
and used tf-idf to normalize
Raw comments (~250,000 out of 1 million) 
Census tallies of houses with internet access 
NLTK 
(natural language processing toolkit) 
to build term-document matrix 
and used tf-idf to normalize 
Found template comments by identifying 
identical rows
Raw comments (~250,000 out of 1 million) 
Census tallies of houses with internet access 
NLTK 
(natural language processing toolkit) 
to build term-document matrix 
and used tf-idf to normalize 
Found template comments by identifying 
identical rows 
Scored sentiment using AFINN-111 
2,477 English words rated between -5 and +5
Raw comments (~250,000 out of 1 million) 
Census tallies of houses with internet access 
NLTK 
(natural language processing toolkit) 
to build term-document matrix 
and used tf-idf to normalize 
Found template comments by identifying 
identical rows 
Scored sentiment using AFINN-111 
2,477 English words rated between -5 and +5 
Front end built using Twitter Bootstrap, AWS, 
and D3
Why might we care about 
template responses?
Working demo
Working demo
Lower engagement 
More positive language 
Higher engagement 
More negative language
Lower engagement 
More positive language 
Higher engagement 
More negative language
Geneviève Smith | Insight Data Science NYC | August 2014
Geneviève Smith | Insight Data Science NYC | August 2014
Geneviève Smith | Insight Data Science NYC | August 2014
Geneviève Smith | Insight Data Science NYC | August 2014
Working demo
Working demo
Working demo
Working demo
Working demo
Working demo
Working demo
Working demo

More Related Content

PDF
Artificial Intelligence, Data and Competition – SCHREPEL – June 2024 OECD dis...
PDF
Storytelling For The Web: Integrate Storytelling in your Design Process
PDF
How to Leverage AI to Boost Employee Wellness - Lydia Di Francesco - SocialHR...
PDF
2024 Trend Updates: What Really Works In SEO & Content Marketing
PPTX
OpenData - potencial v komercnej sfere
PPTX
Koho trapi kvalita v projektoch?
PDF
Internship2015
PPTX
외담대상환청구권 폐지에 뜻을 모아주세요!
Artificial Intelligence, Data and Competition – SCHREPEL – June 2024 OECD dis...
Storytelling For The Web: Integrate Storytelling in your Design Process
How to Leverage AI to Boost Employee Wellness - Lydia Di Francesco - SocialHR...
2024 Trend Updates: What Really Works In SEO & Content Marketing
OpenData - potencial v komercnej sfere
Koho trapi kvalita v projektoch?
Internship2015
외담대상환청구권 폐지에 뜻을 모아주세요!

Recently uploaded (20)

PPT
statistics analysis - topic 3 - describing data visually
PPT
expt-design-lecture-12 hghhgfggjhjd (1).ppt
PPTX
Lesson-01intheselfoflifeofthekennyrogersoftheunderstandoftheunderstanded
PPTX
MBA JAPAN: 2025 the University of Waseda
PPTX
chuitkarjhanbijunsdivndsijvndiucbhsaxnmzsicvjsd
PPTX
SET 1 Compulsory MNH machine learning intro
PDF
Best Data Science Professional Certificates in the USA | IABAC
PPT
lectureusjsjdhdsjjshdshshddhdhddhhd1.ppt
PPTX
Topic 5 Presentation 5 Lesson 5 Corporate Fin
PPT
DU, AIS, Big Data and Data Analytics.ppt
PPTX
Business_Capability_Map_Collection__pptx
PPTX
eGramSWARAJ-PPT Training Module for beginners
PPTX
chrmotography.pptx food anaylysis techni
PPTX
IMPACT OF LANDSLIDE.....................
PDF
Votre score augmente si vous choisissez une catégorie et que vous rédigez une...
PPTX
Phase1_final PPTuwhefoegfohwfoiehfoegg.pptx
PPTX
Crypto_Trading_Beginners.pptxxxxxxxxxxxxxx
PDF
CS3352FOUNDATION OF DATA SCIENCE _1_MAterial.pdf
PPTX
sac 451hinhgsgshssjsjsjheegdggeegegdggddgeg.pptx
PPTX
DS-40-Pre-Engagement and Kickoff deck - v8.0.pptx
statistics analysis - topic 3 - describing data visually
expt-design-lecture-12 hghhgfggjhjd (1).ppt
Lesson-01intheselfoflifeofthekennyrogersoftheunderstandoftheunderstanded
MBA JAPAN: 2025 the University of Waseda
chuitkarjhanbijunsdivndsijvndiucbhsaxnmzsicvjsd
SET 1 Compulsory MNH machine learning intro
Best Data Science Professional Certificates in the USA | IABAC
lectureusjsjdhdsjjshdshshddhdhddhhd1.ppt
Topic 5 Presentation 5 Lesson 5 Corporate Fin
DU, AIS, Big Data and Data Analytics.ppt
Business_Capability_Map_Collection__pptx
eGramSWARAJ-PPT Training Module for beginners
chrmotography.pptx food anaylysis techni
IMPACT OF LANDSLIDE.....................
Votre score augmente si vous choisissez une catégorie et que vous rédigez une...
Phase1_final PPTuwhefoegfohwfoiehfoegg.pptx
Crypto_Trading_Beginners.pptxxxxxxxxxxxxxx
CS3352FOUNDATION OF DATA SCIENCE _1_MAterial.pdf
sac 451hinhgsgshssjsjsjheegdggeegegdggddgeg.pptx
DS-40-Pre-Engagement and Kickoff deck - v8.0.pptx
Ad
Ad

Working demo

  • 1. NeutralOpinion Visualizing how people really feel about net neutrality. ! ! ! ! Geneviève Smith | Insight Data Science NYC | August 2014
  • 6. The FCC doesn’t usually get this massive a response Number of comments/ complaints
  • 7. The FCC doesn’t usually get this massive a response Number of comments/ complaints
  • 9. Raw comments (~250,000 out of 1 million)
  • 10. Raw comments (~250,000 out of 1 million) Census tallies of houses with internet access
  • 11. Raw comments (~250,000 out of 1 million) Census tallies of houses with internet access NLTK (natural language processing toolkit) to build term-document matrix and used tf-idf to normalize
  • 12. Raw comments (~250,000 out of 1 million) Census tallies of houses with internet access NLTK (natural language processing toolkit) to build term-document matrix and used tf-idf to normalize Found template comments by identifying identical rows
  • 13. Raw comments (~250,000 out of 1 million) Census tallies of houses with internet access NLTK (natural language processing toolkit) to build term-document matrix and used tf-idf to normalize Found template comments by identifying identical rows Scored sentiment using AFINN-111 2,477 English words rated between -5 and +5
  • 14. Raw comments (~250,000 out of 1 million) Census tallies of houses with internet access NLTK (natural language processing toolkit) to build term-document matrix and used tf-idf to normalize Found template comments by identifying identical rows Scored sentiment using AFINN-111 2,477 English words rated between -5 and +5 Front end built using Twitter Bootstrap, AWS, and D3
  • 15. Why might we care about template responses?
  • 18. Lower engagement More positive language Higher engagement More negative language
  • 19. Lower engagement More positive language Higher engagement More negative language
  • 20. Geneviève Smith | Insight Data Science NYC | August 2014
  • 21. Geneviève Smith | Insight Data Science NYC | August 2014
  • 22. Geneviève Smith | Insight Data Science NYC | August 2014
  • 23. Geneviève Smith | Insight Data Science NYC | August 2014