Generating Hierarchical Association Rules with the Use of Bayesian Network

Bayesian network (BN) is used to build a model using a probability distribution over a set of variables. The benefit of BN is the compact representation of complex problem domains. Also, BN provides decision making, smooth, consistent and flexible applicability in the complex domains. On the other hand, association rules are based on antecedent and consequent part which have condition attributes and decision attributes respectively. Each association rule has a confidence percentage value which shows the rule effectiveness in terms of probability. Thus, association rules and Bayesian Network can be linked up due to probabilistic approach. Therefore, we focus on association rules to develop an automated data mining technique with the use of Bayesian Network. For this purpose, we suggested Associated Rules Binary Symmetric Matrix using K2 (ARBSM-K2) technique in order to generate Hierarchical Association Rules (HAR). This structure shows the relations among the association rules which show more certainty in the hierarchy after maximizing the probability for each association rule treated as a node. Hence, this hierarchy may open new research dimensions to academia community by refining the existing techniques with the latest techniques to further obtain knowledge in any domain. Thus, the significance of the paper is to utilize it for fraud detection from the start to the end showing all steps in sequence.