Relevant association rule mining from medical dataset using new irrelevant rule elimination technique

Association rule mining (ARM) is an emerging research in data mining. It extracts interesting association or correlation relationship in the large volume of transactions. Apriori based algorithms have two steps. First step is to find the frequent item set from the transactions. Second step is to construct the association rule. If ARM applied with medical dataset, it produces huge quantity of rules; most of these rules are irrelevant to the transaction. These irrelevant rules consume more memory space and misguide the decision making. Here irrelevant rule reduction is important. This paper proposes the n-cross validation technique to reduce association rules which are irrelevant to the transaction set. The proposed approach used partition based approaches are supported to association rule validation. The proposed algorithm called as PVARM (Partition based Validation for Association Rule Mining). The proposed PVARM algorithm is tested with T40I10D100K and heart disease prediction. The performance analysis attempted with Apriori, most frequent rule mining algorithm and non redundant rule mining algorithm to study the efficiency of proposed PVARM. The proposed work reduces large number of irrelevant rules and produces new set of rules with high confidence. It is much use to mine medical relevant rule mining.