Combined Association Rule Mining

This paper proposes an algorithm to discover novel association rules, combined association rules. Compared with conventional association rule, this combined association rule allows users to perform actions directly. Combined association rules are always organized as rule sets, each of which is composed of a number of single combined association rules. These single rules consist of non-actionable attributes, actionable attributes, and class attribute, with the rules in one set sharing the same non-actionable attributes. Thus, for a group of objects having the same non-actionable attributes, the actions corresponding to a preferred class can be performed directly. However, standard association rule mining algorithms encounter many difficulties when applied to combined association rule mining, and hence new algorithms have to be developed for combined association rule mining. In this paper, we will focus on rule generation and interestingness measures in combined association rule mining. In rule generation, the frequent itemsets are discovered among itemset groups to improve efficiency. New interestingness measures are defined to discover more actionable knowledge. In the case study, the proposed algorithm is applied into the field of social security. The combined association rule provides much greater actionable knowledge to business owners and users.