Software quality management: Machine learning for recommendation of regression test suites

Aim/purpose – This study aims to demonstrate machine learning (ML) applications to enhance software development quality management, specifically through optimizing regression test suites. This research aims to demonstrate how ML can predict and prioritize the most relevant regression tests based on software changes and historical testing data, thereby reducing unnecessary testing, assuring software quality, and leading to significant cost savings. Design/methodology/approach – The methodology of this study involves developing and training a ML model using historical data on software modifications and test executions. The model analyzes the data to predict and prioritize the most relevant regression tests for new software builds. This approach is validated through a comparative analysis, whereby the recommendations from the ML model are benchmarked against traditional regression testing methods to evaluate their efficiency and cost-effectiveness. The results demonstrate the practical advantages of integrating ML into software quality management processes. Findings – The conclusions indicate that implementing ML to optimize regression testing has the potential to significantly improve test efficiency and reduce operational costs. The ML model effectively prioritized crucial test cases, reducing the number of unnecessary tests by 29.24% while maintaining the required quality assurance level and focusing efforts on areas with the highest impact. This optimization not only streamlines the testing process but also significantly improves the allocation of resources and cost-effectiveness in software development practices. Research implications/limitations – The research indicated that future studies should adopt more advanced ML algorithms, test these methods on a range of software products, and adopt a more diverse approach to testing. Such an expansion of research may provide better results and a deeper understanding of the role of ML in quality assurance, with the potential to optimize software development processes more broadly. Furthermore, establishing a more robust link between software code and specific tests within the scope of regression tests could enhance the effectiveness of ML-driven recommendations for regression test suites. Originality/value/contribution – Integrating ML into regression testing selection represents a novel approach to the software development process, offering enhanced efficiency and cost savings. This research exemplifies the potential for transforming traditional testing methodologies, thereby making a valuable contribution to the field of software quality assurance. The study demonstrates how advanced technologies can optimize software development processes, reducing costs while maintaining an assured level of software product quality.