AI-driven Outcome Attainment Prediction and Optimization in OBE using Learning Analytics

Outcome-Based Education (OBE) has become a paradigm of engineering and professional education that predicts unequivocal learning results and proven indicators of achievement. However, most institutions are still using traditional, threshold-based, retrospective approaches towards assessing Course Outcomes (COs), Program Outcomes (POs), and Program Specific Outcomes (PSOs). These methods have a lack of predictive abilities and support few features on timely application of academic interventions. The emergence of Learning Management Systems (LMS) and online assessment systems introduces large amounts of educational data, which creates the potential of advanced learning analytics and intelligent decision-making. This paper presents a framework made by AI and used to predict the outcomes achievement and optimize performance in the paradigm of Outcome-Based Education by utilizing the learning analytics. The system proposed combines the academic performance data, attendance history, assessment behaviour, and data on CO-PO mapping in building strong student learning feature vectors. Controlled machine learning models are used to predict the level of CO achievement before the course ends which enables the early detection of risky outcomes and students. The anticipated CO attainments are, in turn, paired with the predicted PO and PSO attainment by the use of weighted mapping strategies that are calibrated in accordance with the accreditation requirements. The suggested framework is scalable, amenable to interpretability, and consistent with accreditation-related standards, which makes it friendly to practical implementation in engineering institutions. This research by shifting the outcome evaluation scheme of retrospective reporting to that of predictive and optimisation-oriented analytics sets the process of implementing the Outcome-Based Education on a more data-based, transparent, and effective paradigm.