An Effective GRU-Based Deep Learning Method for Test Case Prioritization in Continuous Integration Testing

Continuous integrating (CI) testing is a key element in today’s software development process, where automated testing of code changes occurs frequently. Test Case Prioritization (TCP) is a strategy aimed at boosting the performance of CI testing by choosing specific test cases that have a strong likelihood of detecting defects early in each cycle. Since CI testing generates a substantial volume of test execution data, the use of historical test data has become a common strategy in prioritizing test cases. However, many existing methods for test prioritization in CI face challenges: they may struggle with managing extensive test histories, involve numerous parameters that slow down the training process, or being improved for only a few sort of past test cycles. Such limitations can reduce the effectiveness of fault detection in prioritized test suites. In this research work, we developed a Gated Recurrent Unit-driven deep learning (GRU-driven deep learning) model that leverages the principles of regression to prioritize test cases using historical execution data across multiple test cycles. The GRU model learns to identify failed test cases by considering various factors including distance, duration, change in status, last run. The proposed GRU- based deep learning model is evaluated based on two key metrics i.e. time effectiveness and fault effectiveness when compared to some advanced approaches. The efficiency of the suggested methodology was extensively validated through experimentation on two separate datasets named as paint_ control and IOF/ROL. The experimental findings reveal that the GRU-based deep learning model achieves Total Algorithm Running Time (TT) values of 0.02 seconds and 0.01 seconds, along with APFD values of 0.77 and 0.70 regarding the paint_control and IOF/ROL datasets, respectively. These outcomes are notably lower than those obtained by other advanced existing models. It is evident from the result analysis that GRU model outperforms state-of-the-art test prioritization framework based on these two key metrics, and it can efficiently handle large datasets.