Cuckoo Based Rule Induction Algoirthm

摘要

This research work proposes a novel classification data mining algorithm to discover an unordered rule set using cuckoo search with l'evy flight. The proposed algorithm uses Laplace confidence and rule threshold value to measure the quality of a rule. It implements rule convergence test to prevent the redundant rule set. The performance of the proposed algorithm is compared with Ant-Miner and PART algorithm. The results show that (a) cuckoo based classification algorithm is competitive with Ant-Miner and C4.5 with respect to predictive accuracy; and (b) The rule lists discovered by cuckoo based classification algorithm are considerably simpler (smaller) than those discovered by other two algorithms.