Enhanced Multi-Channel Classification Framework for Accurate Student Performance Prediction in Educational Data Mining

Accurate forecasting of student performance plays a crucial role in finding learners at risk and enabling targeted academic support. This research introduces an improved multi-channel classification framework that combines the strengths of Decision Tree (DT), Naïve Bayes (NB), and Support Vector Machine (SVM) algorithms through a weighted voting approach. To further reduce classification errors, a smoothing technique is applied after model aggregation. The study uses academic, demographic, and skill-based data collected from Bachelor of Computer Applications (BCA) students at Saurashtra University. Preprocessing steps—comprising data cleaning, transformation, feature selection, and normalization—were implemented prior to training the models in the R programming environment. Model performance was assessed using 10-fold cross-validation with metrics including accuracy, precision, recall, and F1-score. Experimental findings indicate that the proposed framework achieves an accuracy of 96.39%, significantly outperforming individual classifiers (DT: 78%, NB: 80%, SVM: 83%). The results also highlight the benefit of incorporating subject dependencies and extracurricular achievements in predictive modelling, offering a practical, scalable solution for institutional decision-making.