A Data-Driven Framework Based on Machine Learning Approaches for Restructuring Computer Science Courses

Abstract

This study proposes an approach to machine learning for changing the computer science curriculum at higher education institutions. The approach combines topic modeling, sentiment analysis, clustering, regression, and algorithmic recommendation to derive actionable insights for curriculum design by examining course content, student performance, feedback, and institutional data. The research indicates that this method enhances curriculum relevance, customizes student learning directions, and corresponds with Program Specific Outcomes (PSOs) and Program Educational Objectives (PEOs). Simulation performed using a prototype LMS dashboard confirms the model's viability and applicability. The findings demonstrate how a data-driven framework can proficiently connect educational resources with learner expectations while accomplishing the institutional goals.