A performance based empirical study of the frequent itemset mining algorithms

摘要

Frequent itemset mining is one of the important domains in pattern mining. This deals with mining the frequent itemsets that occur in the dataset. Researches are still an ongoing process in this area. So far many algorithms have been proposed for mining frequent itemsets. Frequent itemsets are mined for framing association rules. Other than framing association rules, mining frequent itemsets leads to effective classification, clustering and predictive analysis. The commonly used algorithms are Apriori, FPGrowth and Eclat. Previously heaps of research works have been made using these three algorithms. Enhancements have also been made to improve the performance. This paper is based on the analysis of these three algorithms with four different datasets having varied number of transactions and size. Comparison is made using the parameters, time taken and the memory used by each algorithm to find the frequent patterns. The result paves way for the future research work in this field.