Data Mining Analysis of Moodle Learning Data and Student Perceptions During and After the Covid-19 Pandemic

Abstract

Listen

Translate article

This study examines the academic performance of students from the 2020 and 2023 cohorts, highlighting differences in activity, attendance, task completion, midterm and final exam scores, and perceptions of educational metrics. A data mining approach was applied to predict students' GPA using Decision Tree, Random Forest, Multinomial Naïve Bayes, and Gaussian Naïve Bayes algorithms. The Gaussian Naïve Bayes model showed the highest accuracy of 0.93 for the 2020 cohort and 0.92 for the 2023 cohort, with the lowest error rate making it the most effective predictor. Feature importance analysis revealed that task completion and exam scores were the most influential factors, while students' perceptions had a lesser impact. The findings suggest that direct academic metrics should be the focus for improving student performance. This study emphasizes the need for further refinement of predictive models and suggests incorporating both academic metrics and student perceptions for a holistic understanding of student performance.