Thursday, April 3, 2014

Can Twitter Predict the Economy Better Than Wall Street Economists?

By Josh Zumbrun

    The researchers who aimed to beat Wall Street by analyzing Twitter didn’t pull it off. Not this week.

    Yesterday we highlighted research out of the University of Michigan that analyzed tweets about job loss and attempted to estimate the number of initial jobless claims that the Labor Department reports every Thursday morning. The method predicted 342,000 people would file claims last week. Wall Street economists had predicted 320,000.

    The report was out this morning and the correct number was 326,000. The economists were closer than the tweets, but the researchers in Michigan aren’t conceding defeat.

    “We were well within the standard of error,’’ jokes Matthew Shapiro, an economist at the University of Michigan who worked on the twitter project, with a team that included survey researchers and computer scientists. “This is an early attempt at something we hope catches on.’’

    Mr. Shapiro and his teammates will stay at it. They’re working on a National Science Foundation grant and collaborating with the U.S. Census Bureau to find more ways to gain useful information on the economy from the enormous amount of data that people voluntarily post to the internet every day.

    “We’re in this moment where there’s a huge amount of data being made available, and there’s opportunities to measure economic data in a way that’s passive – that doesn’t require someone filling out a form or survey,’’ he said. Mr. Shapiro hopes the research will eventually be able to flesh out data series like gross domestic product, that’s only produced quarterly, or to measure concepts that economists and statistics agencies currently don’t track.

    See the original article >>

    No comments:

    Post a Comment

    Follow Us