DATA MINING APPLICATION FOR DETERMINING STUDENT’S ACADEMIC PERFORMANCE

ABSTRACT Predicting student academic performance is a critical tool for identifying at-risk students and improving educational outcomes. This project leverages data mining techniques, employing clustering and classification methods within the Weka framework to analyze attributes such as attendance rate, studyHoursPerWeek, and GPA. By segmenting students into performance categories, the study provides actionable insights to guide targeted academic interventions. The research began with data preprocessing, where missing values were handled, and data was normalized in Excel before importing into Weka. Clustering analysis was conducted using Weka Explorer and Knowledge Flow, while the Random Forest algorithm was implemented for classification. Both training set and cross-validation methods were used to evaluate the reliability and accuracy of the models. Findings revealed that attendance and study habits were the most significant predictors of academic performance. Clustering effectively grouped students into high- and low-performing categories, while Random Forest achieved high classification accuracy, demonstrating its robustness in analyzing complex data. Cross-validation results confirmed the generalizability and reliability of these findings across diverse student datasets. This study highlights the potential of data mining in enhancing academic monitoring systems, offering educators a reliable tool to identify struggling students and provide tailored support. Future research should incorporate larger datasets and additional factors, such as demographics and extracurricular activities, to enhance scalability and predictive accuracy. These advancements would further improve the practical application of machine learning in education.