Machine Learning-Driven Test Case Prioritization Through Data Analytics in Software Testing

چکیده

Software testing plays a critical role in maintaining the dependability and overall quality of software applications. Nevertheless, it often demands significant time and computational resources, especially in unit testing environments. With the continuous expansion of test suites, running every available test case is no longer a practical strategy. Test Case Prioritization (TCP) addresses this challenge by arranging test cases in an order that increases the likelihood of detecting faults earlier while minimizing testing effort and cost. Conventional TCP techniques are often limited in their ability to adapt to the fast-paced nature of current software development practices. To overcome these shortcomings, this research explores the application of machine learning (ML) methods within the TCP framework to improve prioritization effectiveness. In contrast to fixed or rule-based techniques, ML-based approaches can learn patterns from historical test data and make adaptive prioritization decisions. For this purpose, a dataset was generated using several metaheuristic methods, including Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Grey Wolf Optimizer (GWO), and a hybrid GWO model, implemented on the triangle classification benchmark problem. The collected test cases were categorized into five classes according to coverage criteria and then prepared for model training. Four supervised ML classifiers were developed and assessed using stratified 10-fold cross-validation. Their performance was evaluated based on widely used metrics, including accuracy, precision, recall, F1-score, and AUC. Experimental findings showed that the proposed approach reached an accuracy of 97.31% and an F1-score of 92.94%, surpassing traditional TCP methods. These findings indicate that integrating ML into TCP can improve early fault discovery, decrease unnecessary test executions, and offer an automated and scalable approach for more effective unit testing.