Multivariate analysis of Moodle components and grade distribution patterns for adaptive learning environments

Abstract

The increasing sophistication of Learning Management Systems (LMS) has generated vast amounts of educational data with untapped potential for enhancing teaching and learning. While previous research has examined relationships between LMS components and overall student performance, less attention has been paid to how these components influence specific grade distributions. This study employs multivariate analysis techniques to investigate how Moodle components relate to distinct grade categories (A-F) across 985 courses at a Ukrainian pedagogical university. Using multiple regression models, canonical correlation analysis, and cluster analysis, we identify significant relationships between specific Moodle components and grade outcomes. Information components (particularly teacher presence) strongly predict excellence, while Activity x Communication interactions support satisfactory performance. Resource diversity combined with structured activities correlates with reduced failure rates. We further identify four distinct course design clusters with characteristic grade distribution patterns. The findings demonstrate how learning analytics can move beyond descriptive insights toward prescriptive guidance. This nuanced understanding of grade distribution patterns provides practical guidance for tailoring course design to diverse pedagogical objectives and student needs, thus contributing to the emerging field of personalized digital learning environments.