Maintaining high software quality is a major challenge in modern software engineering, as defects in software systems can lead to system failures, security vulnerabilities, and increased maintenance costs. Software defect prediction (SDP) techniques aim to identify fault-prone modules early in the development lifecycle, enabling developers to take preventive actions. This project proposes a hybrid swarm intelligence-based framework that integrates Particle Swarm Optimization (PSO), Genetic Algorithm (GA), and Ant Colony Optimization (ACO) to enhance the performance and accuracy of defect prediction models. The proposed hybrid PSO-GA-ACO framework combines the strengths of multiple optimization techniques to achieve effective feature selection and model optimization. PSO is utilized for its fast convergence and strong global search capability, while GA enhances exploration through evolutionary operations such as selection, crossover, and mutation. ACO contributes by providing a probabilistic search mechanism based on pheromone updating, which helps in identifying optimal feature subsets. The combination of these techniques ensures a balanced trade-off between exploration and exploitation, resulting in improved optimization performance. The framework incorporates comprehensive data preprocessing steps, including handling missing values, normalization, noise reduction, and addressing class imbalance using appropriate resampling techniques. The selected optimal feature subsets are then used to train multiple machine learning classifiers, improving robustness and adaptability across different datasets. The system is evaluated using benchmark software defect datasets, and its performance is measured using metrics such as accuracy, precision, recall, F1-score, and ROC-AUC, along with cross-validation to ensure consistency. Experimental results demonstrate that the hybrid approach significantly outperforms traditional machine learning models and individual optimization methods by achieving higher prediction accuracy, better generalization, and reduced overfitting. Furthermore, the framework improves computational efficiency and scalability, making it suitable for real-world applications. In conclusion, the integration of PSO, GA, and ACO provides an effective and intelligent solution for software defect prediction. This hybrid approach contributes to enhanced software quality assurance practices and supports the development of reliable, efficient, and maintainable software systems.
Akshitha et al. (Wed,) studied this question.