APPLICATION OF DATA MINING FOR PREDICTING STUDENT ACADEMIC PERFORMANCE USING CLASSIFICATION ALGORITHMS

Abstract

The ability to predict student academic performance has become increasingly important for higher education institutions in supporting academic success and reducing dropout rates. Academic performance is influenced by multiple factors such as attendance, previous grades, participation in coursework, and demographic information. However, these data are often underutilized in academic decision-making. Data mining techniques, particularly classification algorithms, provide an effective approach to analyzing historical student data and generating predictive models for future performance. This study applies classification-based data mining methods to predict student academic performance using a dataset containing course grades, attendance records, and cumulative GPA. Several algorithms were tested, including Decision Tree, Naïve Bayes, and k-Nearest Neighbors (k-NN). The methodology involved data preprocessing, feature selection, model training, and performance evaluation using accuracy, precision, recall, and F1-score metrics. The results indicate that classification algorithms are effective in predicting student performance, with the Decision Tree model achieving the highest accuracy at 87%, followed by Naïve Bayes at 82% and k-NN at 80%. These findings demonstrate that data mining can support educational institutions in identifying at-risk students early, enabling timely academic interventions and personalized support. The novelty of this research lies in the comparative analysis of multiple classification algorithms applied to student academic data within the Indonesian higher education context. This study concludes that integrating data mining into academic information systems can significantly enhance decision-making processes and contribute to improved learning outcomes.