In the name of accuracy, Google retools its Flu Trends model

1 Nov 2014 | Author: | No comments yet »

Google Flu Trends Resorts to Actual Data Because It Got It Wrong.

It’s flu season and Google has some news. The new thing is actually traditional data from the Centers for Disease Control and Prevention that is being integrated into the Google flu-tracking model.Remember how excited everyone was about Google Flu Trends last year when it confirmed all of our deepest and darkest fears that we were doomed to a winter of misery? Essentially, it will show you how many searches for particular topics have gone on world wide – but you can also break it down by country or region.

Google has developed a new model for its Flu Trends tool that attempts to predict the intensity of flu season across the world based on the number of searches for certain flu-related terms. Launched in 2009, the original Flu Trends model has performed pretty well, but over-predictions (sometimes significant) over the past couple years have led to a spate of criticism over Google’s methodology.

Companies from Nielsen to large telcos are offering up their customer data to experts who can mine it for clues about public health and social unrest. For example, recently, the Pew research centre ran a worldwide survey of attitudes to homosexuality – unsurprisingly, outside of Europe, a huge number of countries are deeply intolerant of homosexuality. So if you say search, say, “runny nose” or “sore throat” or “runny nose sore throat am I dying,” flu-prevalence in your area would get one more point in its column.

That shouldn’t be surprising: although some researchers criticized Google’s “big data” approach, a truer big data approach would combine data from multiple sources in order to get a more complete picture. Both countries have harsh anti-homosexuality laws on the books, and boast politicians who will happily say being gay is inherently un-Islamic or un-African. The new model is only in effect for the United States this season, but, as Google senior software engineer Christian Stefansen wrote in the blog post announcing it, “We look forward to seeing how the new model performs in 2014/2015 and whether this method could be extended to other countries.” While Google’s model was relatively accurate initially — the company first launched it in 2008 — by the 2012-2013 season, its predictions were far off the mark. But what happens when you have a look at the most common places in the world for common gay porn search terms, as the liberal political magazine Mother Jones did?

To enthusiasts, it seemed so superior to the antiquated method of collecting health data that involved doctors talking to patients, inspecting them and filing reports. Except that, unfortunately for Google, there’s a significant portion of the population (myself included) that consists of wildly neurotic, perfectly healthy hypochondriacs. Of course, it could be argued that this is a chicken and egg phenomenon – without Google, would men who outwardly condemn homosexuality be privately living such online lives?

It did at least give some accurate picture of where flu symptoms were more prevalent, helping scientists predict outbreaks earlier than they might otherwise—to a degree. I’d argue that the desire and urges always existed – it’s just that access to google lets people indulge them, whether pornographically on their own, or by using social media to meet like minded individuals in the flesh.

But to fix the gaping whole in Flu Trends’ logic, Google will start blending their search-acquired data with the CDC’s own, official, old-school variety. It’s perhaps telling that in this excellent BBC piece on secret gay life in Pakistan, one of the men interviewed says “One of the first things I did online, maybe 12 years ago, was type in G – A – Y and hit search. Google has yet to fully disclose exactly how it plans to blend the two, but today’s blog post promises a technical paper explaining the details to be released “soon.” [ via New York Times] David Lazer, a professor of political science and computer science at Northeastern University, and the lead author on the Science article, said it was “great” that Google was overhauling its flu-tracking engine to blend in the C.D.C. data.

Back then I found a group and made contact with 12 people in this city.” We can see a similar phenomenon in Britain with the growth of our own fetish scene. Researchers who discovered the 2012-13 discrepancy asked Google to detail its algorithm with the intention of figuring out where it went wrong, but the search company demurred. Lazer said he and other researchers would like to examine how Google’s service works in detail including seeing at least a subset of the flu-related search terms it monitors — something Google has not yet disclosed.

But Google’s reluctance to disclose more is explained by its concern that Flu Trends runs through the corporate mother ship, the Google search engine. And Google is in a constant cat-and-mouse game with competitors, marketers, advertisers and search consultants trying to figure out the secrets of its search technology.

Before the dawn of the Internet, people who experienced this urge would probably bury their desires deep down, maybe try to slake their needs with hard-to-obtain and dubious mail-order catalogues, or with sex phone lines. While Google doesn’t share the actual query terms in its model — which have numbered between 50 and 300 — Christian Stefansen, the project’s technical lead, said the new algorithm aims to differentiate between people searching for flu information out of curiosity or concern and those who are searching because they have flu symptoms.

Lazer raises a good issue about the terms under which researchers might gain access to the increasingly valuable data sets amassed by private companies. The earlier algorithm may have overshot the CDC’s numbers because it counted people who were searching for news about the flu as an indicator of people who actually had the virus, Lazer says.

The answer, by the way, is always “No”, and it’s always society that’s in the wrong, whether the website is about ladies with an eye for discipline or Ukip membership (if only Google Trends could produce Venn diagrams). For example, The Economist recently ran a piece on the foundering efforts to get cellphone data that could be invaluable for tracking the Ebola epidemic. Sooner or later, whatever your bent, you can insulate yourself among like-minded people, until you think you’re the normal one, and the rest of us are the intolerant. Global Pulse is working with the Universal Postal Service, an organization the tracks global postal travel, to see whether there is a correlation between mail flow and poverty. This area of research is very new, said Dan Kaufman, director of DARPA’s innovation office, in an interview last week at the WSJ.D Live conference in Laguna Beach.

My own feeling is it seems Google is a fascinating window into closed cultures – while people will happily lie to survey takers, politicians and pollsters, they don’t lie when typing things into search engines, looking for their own gratification.

Here you can write a commentary on the recording "In the name of accuracy, Google retools its Flu Trends model".

* Required fields
All the reviews are moderated.
Our partners
Follow us
Contact us
Our contacts

ICQ: 423360519

About this site