Data Mining Human Behavior at MIT

Posted on August 12, 2010 by Jeffrey Newman

MIT researchers are using data mining and large populations of people to study human behavior and recently won Pentagon challenge and $40,000 prize as the first group to report the locations of ten red weather balloons set aloft from secret locations. Thousands of groups signed up for the contest but the winning group lead by Alexander Pentland at MIT’s Human Dynamics Lab created an incentive structure that carried the day. In their system, anyone that reported a balloon to MIT received $2000 if the team won. The person who recruited the finder earned $1000. The person who recruited the recruiter got $500 and so on. In the two days leading up to the competition, the group recruited 100,000 volunteers. Nine hours after the balloon release MIT had sorted the thousands of submissions by computer and won the competitiion. They say they won because they understand the science of incentivizing people to cooperate. In the lab studies are conducted by placing sensors around the subjects neck and install tracking devices in cellphones. They study the movements and evaluate the data to identify facts showing which group members are the most productive. Pentland who has a ph. in artificial intelligence and psychology from Ph.D says he takes data mining into the real world and that the tag and trace methods means they don’t rely on imperfect information culled in social networking sites.