The Latest Innovative Use for Twitter? Predicting Unemployment
Twitter is useful for a lot of things: You can live-tweet your favorite shows, keep track of Justin Bieber’s selfie count, and now, oddly enough, use it to predict the unemployment level at the end of each monthly cycle.
At least researchers at the University of Michigan and Stanford can. A new collaborative study between the schools has spawned a graph that predicts unemployment based on what people are tweeting about work.
The graph’s data is culled in a much simpler way than you might think. The blue line represents seasonally adjusted claims for unemployment insurance, which is information that’s provided by the Department of Labor. The red line is from the University of Michigan’s model that predicts unemployment claims by cataloging the frequency of tweets like “Looks like I have lost my job” or any of the other depressing things you can find when you search for “lost my job” on Twitter.
The graph is now updated weekly, but the initial tool to predict unemployment was based on a massive dataset of 19.3 billion tweets between July 2011 and November 2013. Those tweets were sourced from a list of keywords and phrases that researchers thought would best reflect whether people had lost their jobs (as you can see below).
Most fascinating, however, were the phrases that researchers discovered were unreliable for their search.
“We do not include a search for the words ‘fired,’ ‘benefits,’ and ‘insurance’ alone,” the report explains, “because each was used much more frequently in unrelated contexts (e.g., fired up).”
The combination of the words “lost” and “work” also proved problematic.
“In the case of the phrase ‘lost * work,’ inspection of the matched k-grams clearly indicated nearly universal non-employment related concepts. Many phrases referred to computer problems such as “lost all my work” and “lost my #$% work,” as well as happier references such as ‘lost in my work’ and ‘lost Beethoven work.’ ”
All in all, the team’s report based on Twitter’s data proves that, more and more, social networks are becoming great tools for predicting human behavior. For instance, Facebook can tell you when you’re going to get married. And Google searches can help predict flu trends.
And, if anything, this study absolutely confirms one thing: People are clearly comfortable airing their employment woes for everyone on social media to see them.
[h/t The Washington Post]
Follow Alyssa Bereznak on Twitter or email her here. Follow Yahoo Tech on Facebook right here.
