News & Events > Press Releases > TreeNet® /MART Most Accurate in KDDCup 2004!

TreeNet® /MART Most Accurate in KDDCup 2004!


TreeNet® continues to deliver consistently world class results, topping the field in yet another international data mining competition. In this case one of our clients used a free 3-day evaluation version of TreeNet® to craft the most accurate model (classification accuracy). Overall, taking area under the ROC curve and cross-entropy their TreeNet® submission placed 2nd overall. The cross-entropy winner spent weeks analyzing the data using mostly traditional data exploration methods.

TreeNet® was first place winner in all categories including area under the ROC curve in the Duke/Teradata CRM churn modeling competition of 2002. The current KD2004 win, generated by users with limited TreeNet® experience is especially exciting as it shows that everyday users can also obtain sterling results.

To evaluate TreeNet® on your data please click here.