This work presents an innovative and structured teaching methodology for network traffic analysis using intelligent systems, adapting the CRISP-DM model to the field of cybersecurity. It is the result of extensive experience in supervising and authorizing more than a dozen Bachelor’s and Master’s theses in ICT-related degree programs, which has allowed the identification of common challenges faced by students and the design of a clear, effective framework to address them. The primary goal of this proposal is to provide a practical and replicable resource that supports the development of rigorous and technically sound academic projects, while fostering the acquisition of professional and transversal competencies highly valued in the labor market.

From a pedagogical perspective, the main contribution of this proposal lies in the consolidation of a learning experience that integrates research-based teaching and project-based learning. The methodology guides students through the entire lifecycle of a data science project applied to network security, from the initial understanding of the problem and the collection of heterogeneous data sources —including packet captures, flow data, and system logs— to data preparation, modeling, evaluation, and the deployment of intelligent solutions. This comprehensive approach enhances critical thinking, problem-solving, and autonomous research skills, while also strengthening teamwork and scientific communication competencies.

In terms of teaching innovation, the methodology fosters reflective and evidence-based decision-making by systematizing the key tasks associated with each phase of the process. Students not only learn to apply advanced techniques —such as handling imbalanced datasets, using performance metrics beyond accuracy, or implementing state-of-the-art machine learning and deep learning models— but also to interpret and explain their results using tools like LIME and SHAP. This promotes a deeper understanding of the underlying mechanisms of artificial intelligence, as well as its ethical and operational implications, strengthening both technical expertise and critical awareness.

The educational benefits of this methodology are exemplified through a real-world case study, where the complete process is implemented, from detecting inconsistencies and transforming variables to training and validating models such as Random Forest and XGBoost. This applied dimension immerses students in an authentic research environment, bridging theoretical knowledge with real-world scenarios of network analysis and active cyber defense. Furthermore, it promotes autonomous and meaningful learning processes that prepare students for future academic and professional challenges, aligning the experience with the goals of educational innovation and lifelong learning.

Keywords: Educational methodology, competency-based learning, project-based teaching, data science education.