Department of Economics

2008 Data Mining


A team of SMU graduate students took first place in the 2008 SAS Data Mining Shootout sponsored by SAS, Dow Chemical Company, and the Central Michigan University Research Corporation.  The team members are Stefan Avdjiev, Jayjit Roy, and Manan Roy, all Ph.D. candidates in the Economics department at SMU.  More than 30 teams from universities and colleges across the country competed in the competition.  The announcement was made on October 28 on the second day of the M2008 Data Mining Conference at Caesar’s Palace in Las Vegas.  In February, the teams were given a complex scheduling problem involving a fictitious airline company attempting to maximize customer satisfaction at three airports by properly scheduling hotel rooms and rental cars for customers caught in air traffic delays and cancellations.  Using weather forecasts, among other variables, the teams were to build models to predict whether flights would be cancelled and, if so, pre-reserve a sufficient number of hotel and rental car accommodations for the “stranded” customers of the airline. Each team was given information involving 750,000 arrivals and departures involving the airline upon which to build their predictive models. After the teams submitted their data mining solutions in July, the judges from the sponsoring companies “scored” the proposed solutions on an independent data set of arrivals and departures unavailable to the participants. The teams were then judged on the exposition and accuracy of their proposed solutions. SMU’s solution was chosen as the winning solution in August.  The SMU team then attended, at sponsors’ expense, the Las Vegas meeting for the awarding of the $5,000 winning prize.