Mining multi-dimensional association rules with multiple support constraints

Mining multi-dimensional association rules from a data warehouse has been recognized as an important model in data mining. Earlier work on multi-dimensional association rules confined the minimum supports to be uniformly specified for all items and/or ignored the schema hierarchy between dimensional attributes. This constraint would restrain an expert from discovering some deviations or exceptions that are more interesting but much less supported than general trends. In this paper, we extended the model of mining multi-dimensional association rules to take into account the schema hierarchy and to allow user-specified multiple minimum supports automatically. We also proposed an algorithm, MMS_MDAR, for discovering multi-dimensional frequent itemsets. The experimental result shows our approach is effective.