Comparative Study of Apriori Algorithms for Parallel Mining of Frequent Itemsets

Abstract

Apriori Algorithms are used on very large data sets with high dimensionality. Therefore parallel computing can be applied for mining of association rules. The process of association rule mining consists of finding frequent item sets and generating rules from the frequent item sets. Finding frequent itemsets is more expensive in terms of CPU power and computing resources utilization. Thus majority of parallel apriori algorithms focus on parallelizing the process of frequent item set discovery. The computation of frequent item sets mainly consist of creating the candidates and counting them. The parallel frequent itemsets mining algorithms addresses the issue of distributing the candidates among processors such that their counting and creation is effectively parallelized. This paper presents comparative study of these algorithms.