A Requirement-driven Approach for Designing Data Warehouses

International audienceAbstract. A data warehouse (DW) is an integrated and historised collection of data generally used to make strategic decisions by means of online analytical processing techniques. Most of the existing DW evelopment tools used nowadays in the industry focuses on the structures for data storage, e.g. applying the star or snowflake schema. We believe that DW that better suit the needs of decision makers would be delivered by concentrating more on their requirements. So far, very few approaches have been proposed to elicit DW requirements. This paper proposes a method, called CADWA, for guiding DW analyst in the elicitation of decisions-makers requirements and in their operationalization into a DW model. CADWA shifts the focus from where information from on how they should be structured and why they are needed. To comply with current practice, the approach starts with the elicitation of highlevel requirements, reuses a set of data mart (DM) models and produces a model for the new DW. The paper presents each stage of the CADWA approach, and provides illustrations with an example inspired from a real case