Sci­ence researchers have dis­cov­ered a problem with Google’s Flu Trends system: it’s no longer any good at pre­dicting trends in flu cases.

According to research car­ried out by a team at North­eastern Uni­ver­sity and Har­vard Uni­ver­sity, Google’s Flu Trends (GFT) pre­dic­tion system has over­es­ti­mated the number of influenza cases in the US for 100 of the past 108 weeks — and in Feb­ruary 2013 fore­cast twice as many cases as actu­ally occurred.

A better pre­dic­tion model of the number of cases for the forth­coming week could be more accu­rately gen­er­ated from the number of cases recorded by the US Center for Dis­ease Con­trol (CDC) in the pre­ceding week, the team found.

The dis­covery has led them to warn of “big data hubris” in which organ­i­sa­tions or com­pa­nies give too much weight to analyses which are inher­ently flawed – but whose flaws are not easily revealed except through experience.

Read the article at →