Mar 17, 2015 | Atlanta, GA
Every March a relatively small research page run by a professor from the H. Milton Stewart School of Industrial & Systems Engineering gets a sudden spike in traffic, jumping from a handful of visitors per week to thousands in a few days.
That's because Joel Sokol, Fouts Family Associate Professor, and his colleagues happen to conduct research on one of the biggest sporting events of the year- the NCAA basketball tournament. Their Logistic Regression/Markov Chain (LRMC) ranking system is a computerized model that has had a respectable level of success in picking the men's national basketball champion and overall tournament results over the last several years. During the season, the LRMC uses basic scoreboard data to create a weekly ranking of all 351 Division I NCAA teams. The mathematical formula looks at every game and factors in the margin of victory and where each game is played. The research has been published in several jourals and presented at the MIT Sloan Sports Analytics Conference.
After this year's tournament field was set on March 15, Sokol's team released its bracket, and now sports fans looking for help filling out their own brackets are finding their way to the LRMC website. The model has shown to be more effective than 80 others, including the NCAA's own Ratings Performance Index (RPI).
The 2015 LRMC "Profs' Picks" have top seeds Kentucky and Villanova meeting in the final after each beating two seeds in the Final Four. The model also predicts some early upsets- Texas over Butler and Ohio State over Virginia Commonwealth, as well as a few ninth seeds knocking off eighth seeds.
On the women's side the professors and their model have projected a final consisting of Connecticut and Notre Dame with no major upsets in the opening round.
Sokol is joined on the LRMC team by fellow IGeorgia Tech professors Paul Kvam and George Nemhauser and professor Mark Brown of City College, City University of New York, as well as a dedicated group of undergraduate students.
Click on image(s) to view larger version(s)
Industrial and Systems Engineering