Test Case Prioritization using Clustering Techniques in Software Engineering

Abstract

The increasing complexity of software systems necessitates thorough testing to ensure newlinetheir reliability and effectiveness [1]. Regression testing is a crucial process that newlineverifies whether recent modifications, such as bug fixes and feature additions, impact newlinethe existing functionality of the software. To enhance the efficiency of regression newlinetesting, various approaches are applied, including K-Means selection, minimization, newlineand prioritization [2]. K-Means Prioritization (TCP) is a widely used technique that newlinearranges K-Means based on factors: number of faults detected, fault severity, newlineexecution time, and code coverage. This approach allows critical faults to be detected newlineearlier, improving testing efficiency and reducing potential risks. newlineThe effectiveness of TCP is often measured using key performance indicators such as newlinethe Average Percentage of Faults Detected (APFD), and execution time. Higher newlineAPFD values indicate faster detection of faults, contributing to a more efficient newlinetesting process. Researchers have explored clustering techniques to improve TCP by newlinegrouping K-Means with similar attributes, which helps in systematically prioritizing newlinethem for execution. newlineClustering techniques play a crucial role in enhancing the APFD rate by newlinesystematically grouping K-Means based on key factors: number of faults detected, newlinefault severity, execution time, and code coverage [19]. Research has demonstrated newlinethat clustering algorithms can effectively form distinct groups of K-Means that newlineexhibit similar characteristics. These structured clusters serve as the basis for newlineprioritization, ensuring that K-Means with a higher likelihood of detecting faults are newlineexecuted earlier in the testing cycle. While clustering-based approaches have shown newlinesignificant improvements in K-Means prioritization, challenges remain, including the newlineneed for more robust frameworks and better handling of K-Means attributes. newlineOvercoming the limitations presents an opportunity for further advancements and newlinerefinements in clustering-driven TCP models [22]. newline