It’s the power of social media. The people at PoliticIt say they can accurately predict election winners based on a social media algorithm — and it was an algorithm that had a 92 percent success rate for every major federal race in 2012.
So who are the winners and losers going to be today?
Here’s the methodology they use:
The It Score is a machine learning algorithm that gathers chatter surrounding a political candidate from social, and traditional media sources in order to provide a gauge of their digital influence. It accounts for tone, how people are reacting to a politician, the buzz surrounding the candidate, and what people are saying about the politician. Whatever candidate has the highest It Score will likely win in the election. The machine learning algorithm was trained off of actual primary election results. The original algorithm was predicting at 67% accuracy, but overtime it learned. Currently it is predicting at 92% accuracy, and has a standard error of 5. PoliticIt’s hope is to refine this algorithm so that political candidates can use it to receive real-time feedback on campaign performance.
Based on the polling data, some of the races should be called fairly early, but we might as well take a look at the data for some of today’s candidates and see if the It Scores will be in line with the vote percentages:
