A Machine Learning Approach to Predicting Student Success Through Data Mining of LMS Moodle Activity Data

Learning Management System (LMS) is a digital platform designed to assist learning by disseminating online and blended learning materials. Moodle is one of the LMS platforms widely adopted by educational institutions, companies, and other organizations. At the researcher's university, Moodle is preferred by lecturers and students for facilitating online and blended learning processes, especially since the conclusion of the Covid-19 pandemic. This research aims to investigate the relationship between students' activities recorded in Moodle logs and their academic performance, as well as to evaluate the effectiveness of machine learning models (Random Forest, Support Vector Machine, and Decision Tree) in predicting student success. The sample consisted of activity log data from 376 students obtained via SQL queries from the Moodle database, covering variables such as post/reply scores, forum_read scores, assignment attempts, quiz attempts, number of graded activities, and final grades from the Tech Skills course. Correlation analysis indicated assignment attempt had the strongest positive correlation (0.6) with final grades. Students' final grades were categorized into three classes: good, average, and failed. Among the machine learning methods tested, Random Forest achieved the best predictive performance, with accuracy, precision, recall, and F1 Score all reaching <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{8 1. 6 0 \%}$</tex> in a scenario using <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{1 0 \%}$</tex> testing data and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{9 0 \%}$</tex> training data.