Wikipedia Participation Challenge

This looks like another fun predictive modeling competition hosted at Kaggle: predict how often an Wikipedia user will edit an article over a 5-month period, based on the activities  of 44,000 editors during the first 8 months of 2010. (David Eaves has an interesting post on the genesis of the competition.) Interestingly, while the winner of the competition will receive their $5000 prize based solely on a predictive accuracy score, the goals of the competition are for the Wikimedia Foundation to "quantitively understand what factors determine editing behavior". This makes it a good candidate for using statistical modeling as opposed to machine learning techniques, so that the effect of the explanatatory variables in the data set can be better understood.

Kaggle: Wikipedia's Participation Challenge