Mining Multiple Level Association Rules to Mining Multiple Level Correlations to Discover Complex Patterns

Recently, frequent pattern mining (FPM) has become one of the most popular data mining approaches for various applications such as education, medical, farming, analysis of sale and purchase patterns etc. Apriori algorithm [11] and FP growth algorithm are working efficiently in data mining. These algorithms are typically restricted to a single concept level of hierarchy and uniform support threshold. Sometimes domain database support concept hierarchies that represent the relationships among many different concept levels. In this paper efforts are made to discover items at multiple levels of concept hierarchy. Up till now, a very few concern has been shown to this area. In this study mining multiple levels is explored and extended to mining cross levels in large database on the basis of user specified reduced support threshold constraint. General Terms Identification of complex frequent patterns: from multiple level association rules to cross level association rules