Big Data

Twitter sentiment analysis finds two candidates have never been more controversial-or unpopular

Big data firm Datameer analyzed thousands of tweets in the week preceding the debate, revealing that the two candidates have never been more controversial-or unpopular.

candidatesentiment.png
Image: Datameer

The 2016 election is the most mobile and most social election in American history. Kicking and screaming, the presidential election has caterwauled online, from the primaries to the general election.The first presidential debate of the campaign is nigh, polls are tightening, and voters are listening.

In the week prior to the first presidential debate, analytics firm Datameer analyzed data related to Hillary Clinton and Donald Trump on Twitter and revealed how users feel about each candidate on the eve of the debate. The company develops tools that interpret and visualize large segments of the Twitter firehose for custom verticals, like banking, media, and politics. Sentiment, explained a company spokesperson, was determined by measuring the number of positive words over the total number of both positive and negative words in each tweet. The sample included content from 3,028 users, 5,368 tweets, and 30 days of data.

"The data required some cleansing and preparation, from unifying records to removing stop words," said Datameer Analytics Application Developer Sokhna Vor. "Datameer's text mining capabilities [helped] derive some very interesting results."

SEE: Social Media Policy (Tech Pro Research report)

The resulting tool, US Presidential Candidates, analyzes positive and negative mentions on Twitter, then renders charts that track performance over time. The app is free and presents data with custom Excel-like visualizations. In the week prior to the debate, the application helped visualize information related to how Twitter users are talking about Trump and Clinton. The application scraped data from "tens of thousands" of tweets related to both candidates, explained a company spokesperson, and determined overall sentiment in key states, trending hashtags related to each candidate, and tweet volume and velocity.

US presidential candidate Twitter analysis, according to Datameer:

  • Overall candidate sentiment level (with 100% as the most positive)
    • Clinton: 46
    • Trump: 44
  • Positive and negative terms
    • Clinton: 2,777 positive, 3,233 negative
    • Trump: 1,258 positive, 1,574 negative
  • The top hashtags related to the general election
    • #imwithher, #trump, #crookedhillary, #nevertrump, #maga
  • The top @ Twitter account mentions
    • @foxnews, @potus, @cnn, @mcuban, @barackobama
  • The most engaged states, in order
    • California, Florida, New York, Texas, Illinois
presidentialdebatetwittersentimentanalysis.jpg
Image: Datameer

Read more

About

Dan is a Senior Writer for TechRepublic. He covers cybersecurity and the intersection of technology, politics and government.

Editor's Picks

Free Newsletters, In your Inbox