Investigating Requirements Completeness Metrics for Requirements Schemas Using Requirements Engineering Approach of Data Warehouse: A Formal and Empirical Validation

These days it is essential to maintain the information quality of data warehouse which is used by managers at different levels to take business decisions in organizations. Also, the information quality is assessed by its data models (requirements, conceptual, logical and physical). Various authors have proposed different metrics which were validated formally and empirically to assess the quality of its respective data models. However, no formal and empirical investigation of requirements completeness metrics was witnessed in the literature. Therefore, in this paper, we thoroughly validate the completeness metrics formally and empirically to evaluate the requirements data model quality. As a preliminary step, formal validation using Briand’s framework is carried out on completeness metrics which proves that out of ten metrics, five metrics are size measure, two metrics are cohesion measure, another two metrics are complexity measure and rest one metric is coupling measure. Further, empirical validation includes correlation analysis to ascertain whether completeness metrics are correlated with understandability of requirements schemas. The results illustrate that the eight metrics have positive and strong significant correlation with understandability of requirements schemas. Moreover, linear regression is employed in this study to evaluate the model quality in an objective manner by predicting the understandability of requirements schemas using requirements engineering approach. On the basis of linear regression results, except two metrics, all other eight metrics can build the accurate requirements model.