Generally association rule mining (ARM) algorithms, like the apriori algorithm, initial produce frequent itemsets and afterward, from the frequent itemsets, the association rules that go beyond the minimum confidence threshold. When the...
moreGenerally association rule mining (ARM) algorithms, like the apriori algorithm, initial produce frequent itemsets and afterward, from the frequent itemsets, the association rules that go beyond the minimum confidence threshold. When the data is in large volume, it takes number of scans to generate frequent items.It is a better idea if all the association rules generated directly without generating frequent items and reduce number of scanning of the database. The quality of an association rule cannot only be signified by its support or confidence. There are numerous other metrics existing to determine the quality of an association rule. Then, the concept of ARM can be present as a multi-objective optimization problem in which the objective is to find association rules while optimizing a number of such goodness and quality criteria at the same time. This point of view, evolutionary algorithms have been utilizing extensively for producing association rules. In this paper, in-depth study on various objectives for ARM and some evolutionary algorithms has been done.
