Cyclic Association Rules: Coupling Dimensions And Measures

On-line analytical processing (OLAP) provides tools to explore data cubes in order to extract interesting information. Nevertheless, it cannot offer any explanation of relationships that could exist within data. To achieve this goal, the association rules were performed on data cubes. We focus in this work on a particular class of association rules which is the cyclic association rules. The latter aims to discover rules that occur in user-defined intervals at regular periods. Generally, the generated patterns do not take into consideration the specificities of the multi-dimensional context i.e., the measures and their aggregations. In this paper, we propose a new method of extraction of cyclic association rules from both dimensions and measures. In addition, we redefine the quality metrics of the derived patterns using the summarizability of measures through applying the suitable aggregation functions. To prove the utility of our approach, we undertake an empirical study on a real data warehouse.