EDUCATIONAL COURSE / TUTORIALOUTLETDOTCOM EDUCATIONAL COURSE / TUTORIALOUTLET DOT COM | Page 23

At the center of Gu‘s recent research is Kaggle.com, a crowdsourcing site that challenges IT experts around the world to vie for the best predictive analytics solution to big, knotty questions. Among the big- name users of the site you‘ll find NASA, which used Kaggle to improve the accuracy of imaging dark matter. Allstate Insurance turned to this resource to reduce customer churn and improve estimates on claims costs. Ford sought early detection of driver drowsiness. And, an unnamed, $100plus billion oil and gas company tapped the Kaggle community to better predict oil reserves. What makes research teams successful on Kaggle? Gu crunched some data himself to find out. Specifically, he examined the win ratio of competing teams, as well as the characteristics of the team members and their communication exchanges. ―There are two general types of capabilities involved,‖ Gu explains. ―One is technical capability, or how well people know data mining. The other is social capability.‖ Gu examined the interaction of the two to determine the combination of capabilities likely to be most successful in tackling a contest. ―What we found in the teams that were most successful is that the team leaders have high social capabilities and others on the team have high technical capabilities,‖ he says. ―The leader needs social capabilities to coordinate and solicit knowledge out of highly technical team members.‖ This finding indicates firms might want to look for similar team dynamics, he adds. ―When crowdsourcing within a firm, you must match people with good social capabilities and good technical capabilities. In a competitive environment, this becomes even more crucial.‖ Knowledge management: Breaking the internal barriers Group dynamics also come into play when using social media for company knowledge sharing, according to research conducted by Gu. He teamed with two other scholars to explore what made enterprise social network useful — or not useful — for a large oil and gas firm with some 200 communities sharing information online. Gu says this firm has invested heavily in knowledge sharing. It has staff devoted to the