<p>Addressing student dropout in higher education is a paramount concern for institutions worldwide. This study introduces a novel approach to tackle this challenge by proposing a hybrid predictive model for categorising students into graduates and dropouts. The significance of this research is underscored by the persistent and multifaceted nature of the dropout issue, impacting individuals and society. Leveraging machine learning, the study combines multiple robust algorithms in a model fusion approach, enhancing predictive accuracy, and providing educational institutions with a powerful tool for proactive intervention. The model fusion approach outperforms individual models, excelling in terms of recall, accuracy, precision, and Matthews correlation coefficient. It provides a balanced perspective on dropout prediction. Additionally, this study offers insights into data preprocessing and analysis, guiding institutions in data quality enhancement and model selection. This research introduces a groundbreaking predictive model that empowers educational institutions to proactively identify students at risk of dropping out. Therefore, utilising data-driven approaches, institutions can develop effective retention strategies, foster a supportive learning environment, and ultimately contribute to improved graduation rates and societal well-being. This innovative approach is a significant advancement in addressing student dropout and transforming higher education.</p>