An Improved Apriori Algorithm Established On Probability Matrix

Abstract In this paper the issue of scanning a large database is addressed by using probability matrix to generate the frequent itemset. The method eliminates the candidate having a subset which is not frequent. Currently in Computer Science and Data Mining there are numerous mining algorithms of Associate Rule. Some of the supreme prevalent algorithms are the Apriori which extract frequent itemset from a large database. Although the Apriori algorithm is known to be the best for an Association Rule or Market Basket Analysis there are some challenges such as time wasting in scanning all the items found in the database through repetitive activities and the amount of memory space required as a result of having that large database being scanned. A comparison of the proposed algorithm with Apriori shows that the performance of the Improved Apriori is very promising.