Coupling Metrics for Object-Oriented Data Warehouse Design

Coupling is one of the most important factors influencing complexity and inturn quality of any software artifact. Researchers in the past have proposed different metrics to quantify the structural complexity of data warehouse design. The properties of complexity and size have obtained vast recognition by researchers. Unfortunately, there is little research on metrics to determine the extent of coupling in multidimensional models. None have worked to determine the degree to which the fact and dimension classes are coupled in an object-oriented multidimensional model. This paper puts forward a proposal to determine the extent of aggregation and inheritance coupling among classes in the early artifacts of data warehouse. The proposed coupling metrics take into consideration conformed dimensions which is a significant feature in a data warehouse. Preliminary corroboration of the proposed metrics using a practical framework suggested by Kaner, indicates that the metrics are actually gauging their intended measure. A salient feature of the proposed metrics is that they are accessible during initial phase of the software development life cycle of data warehouse.