A novel greedy approach for Compact Pattern Tree

Organization of transactional data is one of the important steps in Knowledge Discovery. Compact Pattern Tree (CPTree) organization of the data is apt for the FP-Tree, CAN-Tree, CATS-Tree etc., Construction of CPTree has been dealt within two phase method. This paper exploits the transactional data representation in a structured form using one of the data structures for subsequent representation of a transaction. A novel counting based algorithm inspired by Hungarian Method for assignment problem and subsequently the reduct computational method by Johnson Algorithm in Rough Sets has been proposed and demonstrated. The performance evaluation of the proposed method is compared with conventional two phase method on datasets of UCI learning repository.