Coupling Measures and its Impact on Object-Oriented Software Quality

摘要

Metrics measurement is an important means to access the quality of software systems. Software metrics measurement helps in quick and accurate analysis of software quality. Coupling metrics is a kind of software metrics having a significant impact on several qualities attributes of the object-oriented software. In literature, coupling metrics is used to access the software quality attributes like reliability, efficiency, integrity, maintainability, flexibility, portability, reusability and interoperability, etc. of object-oriented software. However, the threshold values for software coupling measures are still not fully defined. This paper investigates the values of seven coupling metrics and explores the impact of coupling metrics values on different quality attributes of the object-oriented software. The Java source code is analyzed to find coupling between different modules of java code. The coupling values of twenty five Java projects are measured to find behavior of coupling values. Using the resulting coupling values of the twenty five projects, percentage threshold values of seven coupling measures are defined. Also the behavior of coupling metrics is plotted using graph. It is found from the results that the percentage threshold value of parameter coupling is in between 24 to 34. Similarly, the threshold percentage of inheritance coupling is between 3 to 13, global coupling is 0 to 7, data abstraction coupling is 9 to 19, import coupling is 18 to 28, export coupling is 19 to 29 and external coupling is 0 to 5. It means parameter coupling, import and export coupling has higher range of values. Inheritance and data abstraction coupling has a medium range of values and global and external coupling have a least range of values. It is also found that the coupling measures have an impact on various quality attributes of the object-oriented software. If the developer could control the values of few important metrics like parameter, inheritance and data abstraction coupling, it will result in improving the overall quality of object-oriented software. Applications: This work is useful for software professionals to maintain level of coupling while developing software. The data set used in this work is totally new and unique.