Bing Predicts sets its sights on US primary elections

The Bing Predicts team and their machine-learning tools are working hard to remind us that Bing is more than a search engine tool and Google competitor. After using their crystal ball right before the NFL Playoffs and the Golden Globes awards, Bing is leveraging the 2016 election as another occasion to perfect their magic.

Before the Iowa Caucus on February 1, Microsoft is warming up to add more data to the Bing Political Index (BPI), which explains where each candidate stands on the key issues. If you’re wondering how Microsoft does its tricks, the team details on a blog post that they’re relying on machine-learning models gathering data from polls, prediction markets, and anonymized aggregated search-engine queries. Of course, forecasts need to be revised often because public opinion is constantly evolving. Of course, Microsoft is proud to remind us that Bing Predicts obtained good results for the 2014 mid-terms, and that they also accurately predicted the ‘no’ vote on the Scottish Referendum.

Primaries predictions and results 012616

Primaries predictions and results

So, what results does the team expect for the February primaries and caucuses? According to an early preview, Donald Trump may win all four Republicans ballots, while Hillary Clinton may win in Iowa, Nevada and South Carolina but lose to Bernie Sanders in New Hampshire.

As always with Bing’s predictions, we’ll see how it goes. In the meantime, if you’re really interested in the upcoming elections, you can get much more data by visiting Bing’s dedicated website.

Share This
Further reading: , , , ,

Do you think machine-learning models can accurately analyze public opinion?