Dynamic Metrics for Object Oriented Designs

Author(s): Sherif Yacoub, Hany Ammar, and Tom Robinson
Venue: Sixth International Software Metrics Symposium, Proceedings.
Date: 1999

Type of Experiement: Case Study



This paper addresses the problem that static analysis techniques aren't able to give the full picture of the quality of an application, so they propose a dynamic analysis metrics suite for OO design. Specifically the suite consists of metrics for dynamic complexity and object coupling based on execution scenarios. These new metrics are then compared to static analysis metrics in an empirical study. The results are as follows:


Number of Associations (NAS) - cannot give distinct indication of coupling
Coupling Between Object Classes (CBO) - Does give indication of coupling
Object Request for Service (OQFS) and Object Response for Service (OPFS) - Indicate the most coupled actors based on the amount of running time.

The dynamic metrics are concluded to give indications of coupling and complexity based on run-time, not only static analysis. Combining static and dynamic metrics will give the full view of a system.