Transformer-based Approach for Predicting Student Admission Category in Higher Education

Deep learning methods are increasingly being utilized in the field of education, particularly in the process of choosing candidates for advanced degree programs. Conventional approaches frequently rely on fundamental statistical methods or uncomplicated learning models that cannot fully grasp intricate patterns in the data. This research suggests a brand-new Transformer-based technique for forecasting student admission outcomes for courses taken after graduation. Using self-attention processes, transformers have shown impressive performance in a number of natural language processing tasks, efficiently capturing long-range links. To leverage its sequential data processing capabilities, we design the Transformer based model to manage the intricacy of student admission prediction tasks. Our approach incorporates structured data sources from extracurricular activities and test outcomes. Compared to baseline methods, the Transformer-based model performs better and more accurately detects minute patterns in the data. The recommended model is tested using a real-time dataset that is generated throughout the admissions process.