2010 UCSD data mining contest

UCSD and FICO are pleased to announce the Seventh Annual University of California San Diego (UCSD) Data Mining Contest. The contest consists of two classification tasks based on data from an e-commerce website.

Contest Tasks

The contest consists of two distinct but related tasks. In each task, contestants are given data about customers and non-customers. The goal is to build a classifier that will recognize customers in unlabeled data. On both tasks, performance is evaluated as AUC (area under the ROC curve).

This year the total prize money available has increased, to $10,000.

Rules: Contestants may enter either or both tasks. They can be alone, or part of a team of up to four people. Contestants may be located anywhere in the world. All current full-time undergraduate and graduate students, and full-time post-doctoral researchers, are eligible to compete. Due to the academic nature of the competition, faculty and persons working for a company may not compete. Teams fall into two categories; teams composed strictly of undergraduates compete in the undergraduate division, and all other teams compete in the graduate division.

Once a day, teams may estimate their performance by submitting their predictions on quiz sets that are of similar construction to the test set. The scores on this set are displayed on the contest leaderboard.

Prizes: $10000 in prizes will be awarded based on task and division, at $2500 per task per division:

• First Place $1250
• Second Place $750
• Third Place $500

Schedule:
Contest starts: July 1, 2010
Final submissions due: August 31, 2010.