Evaluation of object-oriented analysis and design methodologies and investigation of object-oriented design metrics

Date

1995-05

Journal Title

Journal ISSN

Volume Title

Publisher

Texas Tech University

Abstract

This research resuhed in the development of a base framework as a reference model for the classification and subsequent evaluation of a representative subset of existing 00AD methodologies. The fi-amework of the reference model consists of the characteristics and subcharacteristics which need to be measured and the metric definition and rating for each of the subcharacteristics. On the basis of the reference model, h was possible to represent the deviation of each of the methodologies, viz. Booch, Coad and Yourdon; Shlaer and Mellor; Rumbaugh and Wirfs-Brock, fi-om the ideal or classical methodology and also fi'om each other. This classification would guide a project manager to decide as to which methodology to choose for his requirements and also to evaluate a new methodology. The analysis of the survey conducted on the real world designers and developers of the object-oriented appUcations led to a prescription of a minmium set of requirements for an 0 0 methodology. This survey also determined the amount of usage of different methodologies in various applications.The second part of this research resulted in the selection of class coupling as an object-oriented design metric. An empirical validation of this metric was performed, by considering an example object-oriented detailed design. This metric was validated by formulating a hypothesis and analyzing the collected data from the example considered. Class coupling proved to be an indicator of complexity of the design and its maintainability.

Description

Citation