Facilitating Decision-making Process Using a Hierarchical Multi-attribute Model (hmam) in Modeling Relational Data

This work presents a novel approach that is capable of learning relational domain and generating automated hierarchical multi-attribute model (HMAM) to support the development of decision-making. In this paper, we describe the technique of generalizing data in relationaldomainusinggranularity computingasameansofdata summarization toautomateand support the construction of HMAM for decision-making. First, we introduce related works in relational data mining. Then, we introduce the concept of hierarchical multi-attribute model in decision modeling. We proceed by introducing our approach that uses the pattern-based aggregation approach to relational data mining and discuss the pre-processing procedure. Experimental results are presentedbasedon thehepatitis dataset (KDDCUP2005). The results of our analysis show that the proposed HMAMmodel is able to generate rules and the performance of classifier canbe improvedbyadjusting thenumberof clustersgenerated.