Data Mining Cup 2006
Nine Students of our Practical Course on Machine Learning and Data Mining participated in the annual Data Mining Cup. This year's task was to train a classifier to predict whether an item that is on sale at EBay will receive a higher or lower price than comparable items. Three of our students were placed in the top 10 of 189 solutions. Two students were invited to the international Data Mining Days in Chemnitz.
Our best solution (5th place, submitted by Stefan Heinje) was a collaborative effort of all students, the second-best solution (8th, submitted by Florian Nattermann) excelled by dividing the problem into smaller subproblems. Overall, our group exhibited the 2nd-best performance of all participating institutions (cf., e.g., the inofficial university ranking, produced by Daniel Keysers of DFKI).