Graduate Students in Computer Science win 1st Place

Winners of first place in the International Demand Strategy Forecasting Competition sponsored by Caterpillar Logistics' Research and Innovation Group

August 30, 2012

The research team of John Griffith, MS '13, Shua Murtaza, MS '12, Shashwati Ramteke, MS '12, graduate students in computer science,  Dr. Nikolopoulos, project leader and Dr. Ross Fink, faculty consultant, won first place in the International Demand Strategy Forecasting Competition sponsored by Caterpillar Logistics' Research and Innovation Group. The research was supported by a research grant awarded to Dr. Chris Nikolopoulos by Caterpillar for the 2011-2012 academic year.


A number of universities around the world worked on a common research problem and competed to produce the best forecasting system. The winning Bradley team were awarded first place individual trophies and the first place team trophy at an awards ceremony in June.  The second and third place research teams were from the PhD level, tier I institutions, Brunel University in London and Tsinghua University in Beijing, the "MIT of China." All three universities were invited by Caterpillar to Peoria in June and participated in a three-day conference held at Cat Logistics' facility in Morton, where they presented their research and toured Caterpillar facilities.


"The effectiveness of each team's forecasting system was evaluated on how well it optimized such criteria as service level, profit, return on net assets, and inventory turns, when tested on a set of "blind" data provided by Caterpillar" said Nikolopoulos.  "The competing universities were among the best in the world, so it made it especially gratifying for our students and us to win this competition. Our students were able to employ principles and techniques from their Artificial Intelligence, Intelligent Systems and Data Mining classes at Bradley to deliver a far superior forecasting approach, we believe, than the traditional statistical techniques alone."


The project focused on forecasting the need for parts with intermittent demand. The infrequent need for the parts makes inventory control difficult, and traditional statistical forecasting methods are not adequate. This research employs an intelligent logistics approach and may lead to improving companies' ability to forecast unpredictable demand and better control their inventory and costs.