Classification by Aggregating Emerging Patterns

Emerging patterns (EPs) are itemsets whose supports change significantly from one dataet to another. Each EP can be very sharp in differentiating the class membership of a small fraction of instances. By aggregating the differentiating power of EPs, very high classification accuracy can be obtained.
dataset C4.5 CBA CAEP ave sens ave prec
breast-w 96% 96% 98% 98% 98%
ionosphere 92% 92% 92% 91% 92%
iris 95% 93% 100% 100% 100%
mushroom - - 99% 99% 99%
pima 73% 73% 97% 96% 95%
sonar 72% 78% 90% 90% 90%
tic-tac-toe 99% 100% 99% 99% 99%
vehicle 66% 69% 76% 76% 76%
wine 92% 92% 99% 99% 99%