Empirical Validation of Class Diagram Metrics

Author(s): Marcela Genero, Mario Piattini, and Coral Calero
Venue: Empirical Software Engineering Proceedings, International Symposium
Date: 2002

Type of Experiement: Case Study

Quality
3

Summary:

This paper looks into metrics of UML diagrams instead of metrics on source code. Metrics are useful to quantitatively measure the quality of class diagrams in an objective way. Specifically the metrics measure the structural integrity of the class diagrams, which translates into maintainability. The paper finds empirical data based on two controlled experiments evaluating the following metrics:

Number of Classes (NC)
Number of Attributes (NA)
Number of Methods (NM)
Number of Associations (NAssoc)
Number of Aggregation (NAgg)
Number of Dependencies (NDep)
Number of Generalizations (NGen)
Number of Aggregations Hierarchies (NAggH)
Number of Generalizations Hierarchies (NGenH)
Maximum DIT (MaxDIT)
Maximum HAgg (MaxHAgg)

The results showed that all of these metrics are good indicators of class diagram maintainability in relation to Understandability, Analysability, and Modifiability, except NDep. No conclusions could be confirmed regarding the NDep metric.

0