Predicting Student Academic Performance Using Random Forest Regression: A Case Study on LMS Behavioral Data

Abstract

In the evolving landscape of digital education, Learning Management Systems (LMS) have become pivotal in managing student engagement and academic resources. These platforms not only facilitate course delivery but also log extensive behavioral data, including attendance rates, quiz performances, LMS usage time, and forum activities. Leveraging this data, educators and institutions can enhance academic outcomes through predictive analytics. This study investigates the use of Random Forest Regression, a machine learning technique, to predict student final grades based on LMS behavioral data. A synthetic dataset comprising 100 student records was used, each containing features that reflect engagement and performance. The data underwent standard preprocessing procedures including normalization and partitioning into training and testing sets. The Random Forest model was trained and evaluated using Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE) as performance metrics. The model achieved a MAE of 4.57 and RMSE of 5.90, indicating a high level of predictive accuracy. Feature importance analysis revealed that average quiz score and attendance rate were the most significant predictors, followed by LMS time and forum activity. These findings demonstrate the effectiveness of ensemble learning methods in educational settings and support the integration of predictive systems in LMS platforms for real-time academic monitoring. Such systems could provide early alerts for at-risk students and assist educators in designing targeted interventions. This research contributes to the field of Educational Data Mining by validating the practical utility of Random Forest Regression in supporting personalized and data-driven learning strategies.