Abstract

Missing values and incomplete data are a natural phenomenon in real datasets. If the association rules mine incomplete disregard of missing values, mistaken rules are derived. In association rule mining, treatments of missing values and incomplete data are important. This paper proposes novel technique to mine association rule from data with missing values from large voluminous databases. The proposed technique is decomposed into two sub problems: database scrutinizes and rules mining phases. The first phase is used to reexamine transactions which are useful to mine frequent itemset. The second phase is to mine frequent itemset amd construct association rules from valid database. This paper uses Apriori based algorithm in which proposed technique. The proposed technique is tested with synthetic and real T40I10D100K, Mushroom, Chess and Heart disease prediction datasets. Experimental results are shown that the proposed technique outperforms than robust association rule mining (RAR) and Association rules from data with Missing values by Database Partitioning and Merging (AMDPM) algorithm