Theoretical and Empirical Validation of Coupling Metrics for Object-Oriented Data Warehouse Design

Abstract

The conceptual model of a data warehouse can be used to determine its quality during the early stages of design. Metrics have been proposed in the past to quantify the structural complexity of these models. A majority of these metrics focus on the internal quality attributes of size and complexity. Unfortunately, not many measures have been proposed to assess the magnitude of coupling in the data warehouse multidimensional models. Coupling has a significant impact on the complexity and, in turn, quality of these models. In our previous work, we had put forward measures to determine the scope of inheritance and aggregation coupling between classes present in the object-oriented conceptual model of the data warehouse. The proposed measures take conformed dimensions into account, which is a notable feature of the data warehouse. However, the proposed metrics had not been validated. Therefore, the main aim of this study is to corroborate the proposed coupling metrics theoretically against Briand’s property-based framework, as well as empirically, using advanced statistical and machine learning techniques. The results indicate that the metrics are well-founded coupling measures and hence significantly contribute towards the structural complexity of the models which further impacts their understandability.