Empirical Investigation of the Relationship between Interaction Level Metric and Class Defects

Abstract

The object-oriented paradigm has become popular in recent years. In order to better evaluate and control the quality of object-oriented software systems, many object-oriented design metrics proposed to indicate the design properties that influence software quality during design phase. Design metrics proposed should be empirically validated to demonstrate the usefulness. The major focus of our work is to empirically investigate the relationship between class defect count and an existing design complexity metric, Interaction Level (IL). Several open source projects are studied to perform such an investigation. The result shows this metric can be a useful reliability indicator at the class level.