د.ابوعجيلة المصري دغمان2026-06-072026-06-07https://dspace.academy.edu.ly/handle/123456789/2172Modern cybersecurity defense systems face a dual challenge: the increasing complexity of cyberattacks and the "black-box" nature of AI models, which lack transparency. This thesis proposes a Multi-Approach Cyber Threat Prediction Framework that synergizes Machine Learning (ML), Deep Learning (DL), and Explainable AI (XAI) to bridge the gap between high performance detection and operational trust. The methodology involved a rigorous empirical evaluation of multiple algorithms across diverse datasets (CIC-IDS2017, Phishing, and Malware). The findings demonstrate that a "one-size-fits-all" model is insufficient for the evolving threat landscape. Specifically, the Random Forest model emerged as the most robust for malware and network intrusion detection, achieving 99.17% and 99.89% accuracy, respectively, while the Decision Tree model achieved a perfect 100% accuracy for phishing detection. The core contribution of this research lies in the systematic integration of SHAP (SHapley Additive exPlanations), which transforms opaque model predictions into transparent, actionable insights for security analysts. Furthermore, the thesis presents a scalable, cloud-native architectural blueprint, offering a roadmap for transitioning from reactive to proactive defense. The research concludes that the future of resilient cybersecurity depends on the seamless integration of predictive precision, modular scalability, and human-centric interpretability. Keywords: Cyberthreat Prediction, Machine Learning, Deep Learning, Random Forest, SHAP, Malware Detection.Modern cybersecurity defense systems face a dual challenge: the increasing complexity of cyberattacks and the "black-box" nature of AI models, which lack transparency. This thesis proposes a Multi-Approach Cyber Threat Prediction Framework that synergizes Machine Learning (ML), Deep Learning (DL), and Explainable AI (XAI) to bridge the gap between high performance detection and operational trust. The methodology involved a rigorous empirical evaluation of multiple algorithms across diverse datasets (CIC-IDS2017, Phishing, and Malware). The findings demonstrate that a "one-size-fits-all" model is insufficient for the evolving threat landscape. Specifically, the Random Forest model emerged as the most robust for malware and network intrusion detection, achieving 99.17% and 99.89% accuracy, respectively, while the Decision Tree model achieved a perfect 100% accuracy for phishing detection. The core contribution of this research lies in the systematic integration of SHAP (SHapley Additive exPlanations), which transforms opaque model predictions into transparent, actionable insights for security analysts. Furthermore, the thesis presents a scalable, cloud-native architectural blueprint, offering a roadmap for transitioning from reactive to proactive defense. The research concludes that the future of resilient cybersecurity depends on the seamless integration of predictive precision, modular scalability, and human-centric interpretability. Keywords: Cyberthreat Prediction, Machine Learning, Deep Learning, Random Forest, SHAP, Malware Detection.هندسة كهربائية