Cell Phone Data Predicts Movement Patterns 93
azoblue writes "In a study published in Science, researchers examined customer location data culled from cellular service providers. By looking at how customers moved around, the authors of the study found that it may be possible to predict human movement patterns and location up to 93 percent of the time."
This was done last year (Score:5, Informative)
While not to the exactness of this study, this has been done before in May 2009 ( http://www.pbs.org/newshour/updates/science/jan-june09/celldata_05-15.html [pbs.org] ). From the article:
analyzed six months of anonymous cell phone records from more than 100,000 people in a European country, obtained from a European cell phone provider. Those cell phone records gave an approximation of each person's location at the time of each call, because cell phone calls are routed through the nearest cell tower. He and his colleagues found that people tend not to stray far -- almost three quarters of the people stayed mainly within about a 20-mile circle for the entire six months, and nearly half the people rarely strayed outside a six-mile circle. They also tended to go back and forth regularly between only a few locations, such as home and work.
And another attempt on the same idea was done by MIT in July 2005 ( http://yro.slashdot.org/article.pl?sid=05/07/25/1751234 [slashdot.org] ). Difference here was that the percentage was 85%. Not the 93% declared now. From the Wired article:
Eagle's Reality Mining project logged 350,000 hours of data over nine months about the location, proximity, activity and communication of volunteers, and was quickly able to guess whether two people were friends or just co-workers.... Given enough data, Eagle's algorithms were able to predict what people -- especially professors and Media Lab employees -- would do next and be right up to 85 percent of the time.... Eagle used Bluetooth-enabled Nokia 6600 smartphones running custom programs that logged cell-tower information to record the phones' locations. Every five minutes, the phones also scanned the immediate vicinity for other participating phones. Using data gleaned from cell-phone towers and calling information, the system is able to predict, for example, whether someone will go out for the evening based on the volume of calls they made to friends.
Re:This was done last year (Score:4, Informative)
More importantly, people tend to CALL from predictable places. As others have pointed out, most people spend the majority of their time at home and work. But on top of that, these studies only look at where calls are made, not where people actually are. So while I may spend a lot of time out and about on the weekends, I still make the majority of phone calls when I'm at home (not at the movie, shopping, gym, etc..)
The MIT test didn't work based on calls, it used a program that would run every 5 minutes to locate itself based on cell tower information (a low grade GPS). While the test also used calling information, it wasn't for the purpose of figuring out where someone on average would be. Calling information was used to predict whether someone would going out with friends, ect...
Re:Sleep and Work? (Score:3, Informative)
Seeing how 66.67% of the time I am either sleeping at home or at work it shouldn't be too hard to fill the other 27% with commute/grocery shopping.
You're not too far off. I worked at the research wing of a phone company, and I can tell you that "tracking" a person using a cell tower is pretty coarse, even in urban areas. Given that most people go to work on weekdays, I'd say that a lot of your "movement" could be predicted on this level by just predicting your average movement. Add in a weekday/weekend variation, and 93% is hardly surprising.
This isn't even one of those "well duh, in RETROSPECT everything is obvious" studies -- anyone who has ever worked with CDR (mobile phone) data knows that this is pretty obvious even before running the experiments.
And for the people who bring up the MIT Reality Mining experiment, keep in mind that they tracked about 100 *individuals*, all of whom were MIT students with pretty regular routines.