BEHAVIORAL-FACTOR DECISION SUPPORT MODEL FOR PREDICTING ACADEMIC PERFORMANCE

Abstract

Early identification of students at academic risk remains a major concern in higher education because timely intervention can improve retention, learning outcomes, and institutional effectiveness. Educational data mining research has shown that not only grades and test scores but also demographic, family, social, and behavioral variables that can be analyzed for decision support shape student performance. This study developed a behavioral-factor-based decision support model for predicting the academic performance of 178 undergraduate students in the Department of Computer Science, Adamawa State University, Mubi, Nigeria. This study was adapted from the first objective of a 2023 master’s thesis dataset and model framework. The model integrated family support, student motivation, engaged time, extra-curricular participation, study behavior, and related contextual variables. Academic performance was operationalized as the cumulative grade point average, classified into two groups: ≥ 2.0 and ≤ 2.0. After preprocessing and a 70:30 training-testing split, the behavioral-factor framework was evaluated in Python Jupyter Notebook and Weka. The findings showed that Naïve Bayes provided the strongest predictive performance, recording 92.6% accuracy, 89.3% precision, 96.2% recall, and 92.6% F1-score in Python, whereas K-means performed weakly with 38.9% accuracy. This study demonstrates that a behavioral-factor model can serve as a practical decision-support mechanism for the early identification of academically vulnerable students. This study recommends the institutional adoption of behavioral monitoring and SPA for student support and academic planning.