STATISTICAL BASED REFILTERIZATION OF POSITIVE AND NEGATIVE ASSOCIATION RULES FROM TRANSACTIONAL DATABASE

Abstract

Association rules are normally generated from frequent itemsets but negative association rules can also be generated from infrequent itemsets as these negative rules are also beneficial to the user. In this paper, the complementary problem i.e. negative association rule mining is considered along with positive association rule mining. Statistical measures are used to find the interested rules. Even though the rules are interested they are not useful to the user then there is no use of such rules. so in this paper it has been overcome with refilterization mechanism The proposed method utilizes Item based Bit Pattern and strong pruning mechanism at item level as well as rule level. Interest based measures are incorporated in this proposed method to reduce uninteresting positive and negative rules.