ABSTRACT

Mining of association rules has become vital in organizations for decision making. The principle of data mining is better to use complicative primitive patterns and simple logical combination than simple primitive patterns and complex logical form. This paper overviews the concept of temporal database encoding, association rules mining. It proposes an innovative approach of data mining to reduce the size of the main database by an encoding method which in turn reduces the memory required. The use of the anti-Apriori algorithm reduces the number of scans over the database. The Apriori family of algorithms is applied on the encoded temporal database and their performances are compared. Also an important method on how to track the association rules that change with time is focused. This method involves initial decomposition of the problem. Later the changing association rules are tracked by dividing the time into smaller intervals and observing the changes in the itemsets obtained in each such interval. Thus the results obtained are lower complexities of computations involved, time and space with effective identification of changing association rules resulting in good decisions making. This helps in formalizing the database metrics in a better way.