AI and Machine Learning in Software Testing: Bridging the Gap Between Defect Prediction, Test Automation, and Continuous Integration

Software testing is an essential process in software engineering that assures the reliability and functionality of software applications by recognising defects early in the development cycle. Bugs, often known as software defects are challenges that cause software to function improperly or fail to meet its desired specifications. Traditional defect detection techniques include static analysis and manual testing, which are laborious and prone to human errors, which could miss defects pertaining to runtime behavior. In recent years, test automation has become an increasingly common method to enhance the testing process. It integrates with Continuous Integration (CI) pipelines to ensure that errors are continuously detected when new code is released. By employing the PROMISE dataset and a gradient boosting technique, Light GBM, the proposed research improves the accuracy of software defect prediction (SDP). The model employs gradually larger and more varied datasets to increase the model efficiency through a model retraining procedure. Retraining increased precision from 0.93 to 0.98, recall from 0.91 to 0.97, and F1-score from 0.92 to 0.97, according to measures that were assessed before and after retraining. The findings show that the LightGBM model greatly improves problem identification and yields more accurate predictions when used in conjunction with automated test cycles. This optimizes testing procedures and shortens the time-to-market for software releases.