Predictive analytics in education- enhancing student achievement through machine learning

This study investigates the application of predictive analytics and machine learning models to enhance student achievement in educational settings. The experiment involved a dataset of 24,005 student records collected from institutional academic records at Wollo University and the Kombolcha Institute of Technology, spanning the years 2017–2022. The data were systematically gathered from student demographic information, academic performance metrics, and contextual features such as studied credits, entrance results, and number of previous attempts. Unlike prior works, this study proposes a novel hybrid architecture that combines Convolutional Neural Networks (CNNs) and Random Forests with XGBoost as a meta-learner, achieving superior accuracy (88 %) compared to individual models such as Random Forest (85 %). Accuracy and other performance metrics (precision, recall, F1-score, and AUC-ROC) were calculated using a hold-out validation approach, with 80 % of the data used for training and 20 % for testing. This architecture effectively captures complex feature interactions and provides actionable insights for educators. Additionally, key predictive factors such as studied credits, entrance results, and regional differences were identified, offering a comprehensive understanding of student performance. The study addresses gaps in feature diversity and demonstrates the applicability of hybrid models in educational settings, paving the way for targeted interventions and improved resource allocation.