The earlier at-risk students can be detected, the better students’ performance and lower the dropout rate can be achieved. The goal of this research is to develop an explain-able machine learning framework for early predicting students’ academic performance on Open University Learning Analytics Dataset (OULAD). It employs the information available from early-stage features such as demographics, students’ engagement and initial assignment grades to make predictions. Several ML models were applied and a Random Forest classifier obtained the best prediction performances over other methods concerning accuracy, precision, recall and F1-score. The proposed system added the technique of SHAP (SHapley Additive Explanations) for both global and local interpretation of model predictions and it also provides explanations through a user-friendly dashboard. The outcomes indicate that this proposed framework provides an excellent and efficient solution to make decisions based on data that yields a timely and insightful system with accurate predictions and clear explanations.

Index Terms—Early Student Performance Prediction, Machine Learning, Random Forest, Explainable Artificial Intelligence (XAI), SHAP, Educational Data Mining, OULAD Dataset, Pre-dictive Analytics, Student Engagement, Decision Support System