An Empirical Study on Object-Oriented Metrics

The objective of this study is the investigation of the correlation between object-oriented design metrics and the likelihood of the occurrence of object-oriented faults. Such a relationship, if identified, can be utilized to select effective testing techniques that take the characteristics of the program under test into account. Our empirical study was conducted on three industrial real-time systems that contain a number of natural faults reported for the past three years. The faults found in these three systems are classified into three types: object-oriented faults, object management faults and traditional faults. The object-oriented design metrics suite proposed by Chidamber and Kemerer is validated using these faults. Moreover, we propose a set of new metrics that can serve as an indicator of how strongly object-oriented a program is, so that the decision to adopt object-oriented testing techniques can be made, to achieve more reliable testing and yet minimize redundant testing efforts.