MACHINE LEARNING-BASED PREDICTIVE MODELING OF STUDENT ACADEMIC PERFORMANCE

Towards predicting university student academic success five machine learning algorithms (Logistic Regression, Decision Trees, Random Forest, Support Vector Machines, and Neural Networks) will be evaluated for their performance effectiveness in this study. The research investigates which predictive modelling technique provides maximum reliability for data-driven choices within higher education institutions. The research evaluated different machine learning methods when they analyzed educational datasets. The evaluation of models utilized accuracy precision along with specificity and the F1 score for metric assessment. A systematic testing method was used during model training and testing to establish reliable results for every algorithm. Neural Networks produced the most effective results with 69.83% accuracy and 80.29% F1 score, yet Random Forest achieved 69.27% accuracy combined with 80.09% F1 score. The accuracy measures of Support Vector Machines and Logistic Regression reached 69.27%, but Decision Trees produced 65.88% accuracy. Educational data analysis benefited the most from complex models when these models surpassed simpler algorithms in identifying complex associations. The research establishes an inclusive evaluation of different machine learning applications on educational data which addresses a literature gap about predicting methods in academic environments. Based evidence supports institutions wishing to implement predictive analytics systems for student performance monitoring.