Generative Artificial Intelligence (GenAI) techniques have received massive and widespread industry-disrupting attention because they can create synthetic data, images, text, and other types of content that look very close to real-world examples. Unlike traditional AI that mainly focuses on classification or prediction, GenAI models learn and copy data distributions. This allows new applications such as content creation, data augmentation, and training of machine learning models. In this survey, we organise the discussion around two major themes: general strategies in GenAI and domain-specific adaptation. We develop a mapping approach to help researchers and practitioners find suitable GenAI techniques for different domains. For each domain, we provide detailed discussions about data considerations, real-world case studies, the use of domain-related tools and frameworks, and practical guidance. We also review domain-specific challenges and opportunities to support future research and application work. Our findings give a full view of the changing GenAI field and suggest ways to build more effective, flexible, and strong generative solutions.

Student dropout rates continue to present major difficulties for educational institutions, leading to academic, operational, and financial impacts. Educational Data Mining (EDM) methods, particularly those combining clustering techniques with predictive models, have demonstrated potential in identifying at-risk students early and accurately. This systematic review explores how cluster-based prediction models have been applied in educational contexts to enhance student performance prediction. A total of sixty-one relevant studies published between 2010 and 2025 were selected and analysed using PRISMA guidelines. The review focuses on the clustering techniques used, how these are integrated with predictive models, and what types of student data are involved. Key findings show that cluster-based models help capture behavioural and academic differences among students, which enables educational institutions to provide more adaptable support. The review also highlights challenges related to generalisability, scalability, and ethical concerns, especially when applying models across different institutions or datasets. The main contribution of this study is the identification of how clustering can be used not only to segment student populations but also to improve prediction accuracy by tailoring models to each subgroup. This review contributes to the literature by emphasising the practical benefits of cluster-based predictive modelling and providing clear directions for further studies aimed at reducing student dropout through targeted support.

Accurately identifying students at risk of failing or dropping out has become a key focus in higher education, where early intervention is essential for improving academic outcomes and retention. While many predictive models have been developed, their generalisability across different courses and learning contexts remains limited. This study explores whether clustering techniques can enhance the performance and adaptability of predictive models in educational data mining. Using the Open University Learning Analytics Dataset (OULAD), K-means clustering was applied to cluster courses based on behavioural, demographic, and academic features. Five machine learning models—logistic regression, random forest, neural networks, LightGBM, and support vector classifier—were trained both on the full dataset as a benchmark, and within each cluster. Cluster-based models consistently outperformed baseline models. The average AUC improved from 0.80 to 0.86 (+7.3%), and mean F1-scores increased from 0.66 (baseline) to 0.78 (best cluster-based model). The highest cluster-specific F1-score achieved was 0.88 using LightGBM. This improvement result from grouping courses with similar behavioural and demographic patterns, allowing models to learn context-specific risk factors more effectively. Clustering also uncovered distinct engagement profiles, such as consistently low activity, sporadic participation, and sustained high interaction, that enhanced model interpretability and allowed for more targeted identification of at-risk students. These findings highlight how segmenting courses by shared behavioural and demographic features can significantly improve both the performance and applicability of predictive models in diverse educational environments.