Ripple Effect Identification in Software Applications

Abstract

Changes are made frequently in software to incorporate new requirements. The changes made to one class are not limited to that particular class, but they also affect other entities. Early identification of these change prone entities is very essential for minimizing future faults in the software applications. Thus, it is very important to develop quality models for identifying the ripple effect of changed classes to effectively utilize the limited resources during the software development lifecycle. Association rule mining is a popular approach suggested in literature, but a major limitation of this approach is its inability to generate recommendations in case of new addition of classes. This article suggests the development of prediction model using learning techniques to overcome this limitation. The authors evaluate the performance of thirteen statistical, ML, and search-based techniques using eight open source software applications in this work. The findings of this study are promising and support the application of SBT and ML techniques for ripple effect identification.