Search

The task of mining association rules consists of two main steps. The first involves finding the set of all frequent itemsets. The second step involves testing and generating all high confidence rules among itemsets. This paper presents a comparative study of association rules algorithms on large databases. Five algorithms have been chosen for this comparative study. The Apriori, Close, FP-growth, Top-k rules, and TNR algorithms have been chosen because these are the most commonly used in the literature. Moreover, these algorithms differ in the number of dataset scanning which affects the performance. In addition, some of these algorithms generate redundant association rules while others don't. All these algorithms are implemented and compared on different datasets. Experimental results show that the FPGrowth algorithm has the best performance, while the TNR algorithm has the best generated non-redundant association rules, and the Top-k rules algorithm has the best performance when the minimum confidence is high. [ABSTRACT FROM AUTHOR]