Data Mining Cup 2007
As in previous years, as part of our Practical Course on Machine Learning and Data Mining, we participated in the annual Data Mining Cup. This year, one of our students, Stefan Appel, finished in third place (among 248 submitted solutions from 688 registered students of 159 universities from 40 different countries). In total, three of our students finished among the top 10 (and were invited to participate in the international Data Mining Days in Leipzig), and five students finished among the top 20.
This year's task was to optimize the costs of two coupon promotions by training a classifier that is able to make cost-optimal predictions for whether customers would return the promotional coupons, based on their behavior in previous promotional activities. Besides making solid use of the provided cost information, Stefan Appel's award-winning solution uncovered hidden dependencies within the order of the customers in the provided data sets.