Towards a Hybrid Model: Evaluating Classification and Regression Models for Student Academic Outcomes

This study develops and evaluates a hybrid machine learning model that combines Random Forest and Gradient Boosting to predict undergraduate student outcomes more effectively. Using data from over 8,941 student records, including demographic and academic records, the approach first identifies Random Forest as the most accurate classifier for predicting graduation probability, achieving an accuracy of 87.82% and approximately 90% after fine-tuning. Gradient Boosting emerges as the top regression model for estimating average GWA, with an R² of 0.8402 that improves to 0.8550 after tuning. By integrating Random Forest’s classification capability with Gradient Boosting’s regression precision, the resulting hybrid model enables simultaneous prediction of graduation likelihood and academic performance. This approach offers comprehensive recommendations for academic advising support, while also highlighting the importance of progressive machine learning techniques in enhancing decision-making.