Empirical Analysis of Object-Oriented Coupling and Cohesion Measures in Determining the Quality of Open Source Systems

Normi Sham Awang Abu Bakar

Keywords

Software metrics, object-oriented design, software quality, open source systems

Abstract

Design metrics play an important role to help developers understand the software design aspects which later can be used to improve software quality. This paper provides empirical evidence supporting the roles of object-oriented (OO) coupling and cohesion metrics, using a subset of Chidamber and Kemerer OO metrics suite, in predicting the post-delivery defects. The findings, based on various open source systems developed in Java, indicate that, a coupling measure, Coupling between object classes (CBO) is very highly significant in predicting defects while Lack of cohesion metric (LCOM), as a cohesion measure is not significant. Even after controlling for size, CBO metric is still significant as a predictor for post-delivery defects.

Important Links:



Go Back