AN IMPLEMENTATION OF ENHANCED APRIORI ALGORITHM FOR RULE GENERATION USING WEB LOG MINING

Abstract

Apriori algorithm is classical algorithm for finding frequent sets and rules. It has two drawbacks i.e. it generate large candidate item-sets and second it require multiple time scanning process of entire database. Many algorithms are proposed on improvement of Apriori algorithm that tries to solve problem of classical algorithm. To generate association rule from data-sets in a manageable size, it is an important to solve the above two problem in a single method. In this paper, we use the advantage of previous algorithms that are improvement of classical Apriori algorithm and design that solve the problem of classical Apriori algorithm. The algorithm requires only two scan of data-sets or database and generates all frequent item-sets and association rules according to user define support and confidence values. The concept used in the algorithm is the elimination process after generating one item frequent set by scan of data-sets first time, which removes the data-items from transaction which are not present in one item frequent set list. This step provides the benefit to eliminate the large infrequent candidate generation. At the end we demonstrate the our approach that produce the association rule mining that requires less time and less infrequent candidate generation than other previous algorithm and the comparison result of the algorithms shows that our algorithm performs batter with respect to complexity analysis that require for execution.