Google is a pioneer in using big data to highlight trends and reveal interesting information, but even its algorithms can go wrong sometimes. Google Flu Trends has been a useful tool for determining how much the flu virus spreads each season, until now.
The idea is that as more people get the flu, they start Googling information on it. The more searches, the more sick people, so Google is able to provide fairly accurate real-time data, something not available to doctors and researchers in any other way.
But the method is not foolproof, circumstances this year led to Google overestimating the number of sick people by as much as 200 percent.
The flu season in the US got off to an early start this season and peaked around Christmas, much earlier than usual. There was a higher mortality rate as well.
This contributed to lot more media attention and reports on the flu which, in turn, led a lot more people than usual to search for info on the flu.
This influx of searches tripped off Google's measurements resulting in wildly wrong estimates. Of course, with this new data Google is able to make adjustments to the algorithm, but this shows that it's still early days in this field.