Enhancing Fault Prediction in CBSE Systems Using Deep Learning and Hybrid MVO-PSO with Shapiro-Wilk Mechanism

Fault prediction is very vital for software dependability and performance in CBSE systems. Choosing ideal qualities for defect prediction, however, is a challenging dance. This paper investigates the relevance of deep learning in feature selection for defect prediction in CBSE systems using a hybrid MVO-PSO approach along with the Shapiro-Wilk technique. By integrating deep learning models with an optimal feature selection framework, the suggested approach reduces computing cost and increases prediction accuracy. Though the Shapiro-Wilk approach statistically confirms feature relevance, MVO-PSO is utilized to select the most relevant features so attaining the best compromise between exploration and exploitation. The performance of the proposed model on benchmark CBSE datasets is evaluated in contrast to traditional machine learning techniques. Experimental results indicate that the hybrid deep learning architecture significantly improves fault prediction accuracy measures while preserving computational economy. By means of comprehensive study on intelligent feature selection methods for defect prediction in CBSE systems, this work enhances the dependability and software quality. With 98.8% accuracy, 98.2% precision, 98.7% recall, and 98.4% F1-score, experimental findings show that the suggested Hybrid MVO-PSO + Shapiro-Wilk model performs much better. Furthermore, it shortens testing time to 24.8 seconds and training time to 105.2 seconds. This validates the model's efficiency in improving prediction accuracy and reducing computing cost in comparison to more traditional approaches.