New SmogFarm Technology Reliably Predicts Gallup Mood Index Results One Day in Advance
SANTA BARBARA, Calif., Jan. 17, 2012 /PRNewswire/ — A research study by SmogFarm demonstrates that its new information processing system can use processed data from Twitter(TM) to predict the results of The Gallup Organization(TM) mood index one day in advance of index data posting.
“Our emotions tracking system provides a method for producing a mood index very similar to Gallup’s but more quickly, reliably and for a fraction of the cost,” said James Crane-Baker, co-founder of SmogFarm (http://smogfarm.com).
Gallup tracks Americans’ views on politics, economics and wellbeing by conducting 1,000 or more nightly interviews. The SmogFarm tracking system acquires data by automatically monitoring millions of tweets per day, processing that information and producing near real-time assessments of values for 12 emotions at a given hour or day using natural language processing methods.
“We created an experiment around the hypothesis that our SmogFarm system can predict various outputs from sentiment engines and indices without requiring costly and time- consuming polling and analysis operations,” said Crane-Baker. The experiment analyzed outputs from the Daily Mood Index, one of the sentiment indices in the Gallup- Healthways Well-Being family of indices. The index reflects whether interviewees are experiencing happiness and enjoyment versus stress and worry.
Crane-Baker and his team selected a standard Classification and Regression Tree algorithm to build a predictive model for the Daily Mood Index, using SmogFarm emotions (and some permutations) as predictor variables. The SmogFarm Daily Mood Index was correlated to the Gallup mood indices for the period of August 12, 2011 to December 16, 2011.
The study concluded that SmogFarm emotion indices tracked the Gallup Daily Mood Index for happiness or enjoyment very closely. “Our data is available one day earlier, so SmogFarm is tracking the emotions of the world in near real-time as they are being formed in the experiences of the Gallup interviewees,” said Crane-Baker.
“Other sentiment analysis tools are largely limited to simple positive, negative or neutral evaluations of sentiment, but our SmogFarm NLP tracking and scoring engine generates minute-by-minute values for a broad range of 12 core emotions, and is based on the science of emotion psychology,” said SmogFarm clinical research psychologist Dr. Erin Olivo. The emotions assessed are anger, confidence, contentment, disgust, envy, excitement, fear, happiness, love, relief, sadness and shame.
Crane-Baker concluded, “We’re advancing the concept of what we call ‘crowd psychology.’ The success of this experiment suggests that the SmogFarm system could be used to gauge general attitudes, beliefs and ideas in the general U.S. population for anything from election results to box office receipts and more. What’s even more exciting is that research shows that people sometimes have emotional reactions before an event occurs. That’s a future goal: gauging this presentiment.”
The full study is available at http://tinyurl.com/6o4bns4.
SOURCE SmogFarm LLC