Georgia Tech researchers, relying on a computerized ranking system they say historically has proven more accurate than the NCAA's own system, predict the University of Kentucky's men's basketball team will win this year's tournament.
The Logistic Regression Markov Chain (LRMC) system has Kentucky beating Ohio State University for the championship, with Michigan State and Kansas being the other Final Four teams. (Yes, the predictions were made before the tournament started.)
The computational system spits out its results after chewing on scoreboard results, home court advantage, margin of victory and other factors.
"Kentucky is the likely champion because they've won almost all their games," said Joel Sokol, operations research professor at Georgia Tech who developed LRMC along with colleagues, in a statement. "They've won by convincing margins at home and on the road against very good teams, and they've done it all against a strong schedule, including Kansas, North Carolina, Indiana and Florida."
Last year, Georgia Tech researchers predicted Ohio State would win, and didn't even have eventual winner University of Connecticut in its Final Four.
As always seems to be the case with the Georgia Tech predictions, LRMC doesn't seem to go out on too much of a limb, picking two No. 1 seeds and two No. 2 seeds for the Final Four. And that's probably a good thing since its insights into teams that might pull off first round upsets and regarding Cinderella teams that might make it into the Sweet Sixteen round were mainly off the mark (only North Carolina State, of its upset picks, has survived).
Georgia Tech has been publicly releasing its predictions for the past few years, though says that since 2003 the LRMC has corrected called the results of more NCAA tournament games than other ranking systems such as Sagarin's predictor, Pomeroy's ranking, Las Vegas Favorite and the NCAA's RPI.
Sadly for Georgia Tech, its team once again did not make the tournament for the second straight year.
Bob Brown tracks network research in his Alpha Doggs blog and Facebook page, as well on Twitter and Google +.
Read more about data center in Network World's Data Center section.