Empirical Evaluation of the Impact of Refactoring on Internal Quality Attributes

Abstract

It is commonly believed that refactoring increases software quality. This paper presents the validation of belief about refactoring effect on internal software quality attributes, notably coupling. High coupling between classes is one of the smells considered dangerous by developers because of its impact, decreasing code maintainability. Some developers believe that refactoring as a method that can improve maintainability by reducing coupling in the source code. Evaluation of the effect of refactoring is done in the experiment way. When data collection from the experiment is done, then statistical analysis could be conducted to see whether refactoring can improve coupling or not. The results show that proper refactoring scenarios significantly reduce the coupling metric in source code, where 86% of samples have zero value of DCH, showing an apparent gain from refactoring. However, developers should pay attention to the code's size as refactoring results in increasing the code's size, especially in NOM, which has increased by 1.81%.