Data mining techniques application for prediction in OLAP cube

Abstract

Data warehouses represent collections of data organized to support a process of decision support, and provide an appropriate solution for managing large volumes of data. OLAP online analytics is a technology that complements data warehouses to make data usable and understandable by users, by providing tools for visualization, exploration, and navigation of data-cubes. On the other hand, data mining allows the extraction of knowledge from data with different methods of description, classification, explanation and prediction. As part of this work, we propose new ways to improve existing approaches in the process of decision support. In the continuity of the work treating the coupling between the online analysis and data mining to integrate prediction into OLAP, an approach based on automatic learning with Clustering is proposed in order to partition an initial data cube into dense sub-cubes that could serve as a learning set to build a prediction model. The technique of data mining by regression trees is then applied for each sub-cube to predict the value of a cell.