This thesis uses different AI techniques to examine predictive modeling for detecting Distributed Denial of Service (DDoS) attacks in 5G networks. The study explores traditional supervised models, unsupervised anomaly detection models, and sequential methods. It evaluates them based on accuracy and their ability to minimize false positives and false negatives. The study trains and tests the models on a 5G network traffic dataset. It applies SHAP and LIME to interpret the models and improve transparency, making their decisions easier to understand. The results show that the traditional models perform better than the others, with Gradient Boosting performing the best. This strong performance likely results from the dataset’s clear and separable fea-tures, which let Gradient Boosting create accurate splits without needing deeplearning complexity. The study emphasizes that both model performance and interpretability matter for building trust in AI-based cybersecurity. Although models like Decision Trees are interpretable by design, this study still applies SHAP and LIME to provide consistent, detailed explanations across both simple and complex models.