Proposing and empirically validating change impact analysis metrics

This PhD project focuses on proposing methods and tools for assessing change proneness and instability. These two quality attributes are of great importance for performing efficient change impact analysis, which is vital during the software maintenance phase. Change impact analysis is important both before and after the application of the change. The project takes into account existing literature, in which we have identified specific limitations: (a) lack of metrics at the architectural design and requirements phases, (b) lack of metrics accuracy at the implementation and detailed-design phases, and (c) lack of tools that can automate the process of calculating them. Considering the aforementioned limitations, the project proposes four novel metrics for quantifying change proneness and instability at the level of requirements, architectural design and source-code. All metrics have been rigorously validated using empirical practices, such as case studies on open-source and industrial software, as well as mathematical proofs. Empirical validation has been performed based on the IEEE Standard on Software Measurement. The results of all studies suggested that the proposed metrics are the most accurate predictors of change proneness and instability. Based on the findings several actionable results for both practitioners and researchers have emerged. To support the applicability of the proposed methods in practice, all metrics are accompanied by a corresponding tool that automates their calculation for java projects.