DECODING LEARNERS: EXPLAINABLE AI FOR INTELLIGENT STUDENT PROFILING IN ONLINE CODING PLATFORMS

Abstract

Online Judge Systems (OJS) have become essential in assessing programming skills by automatically evaluating students’ code submissions. However, most existing systems focus solely on quantitative metrics such as accuracy, execution time, and memory usage—offering limited insights into students’ learning behaviors. This paper presents “Decoding Learners,” an Explainable Artificial Intelligence (XAI)-based framework designed to profile students intelligently within online coding platforms. By leveraging interpretable machine learning models such as SHAP (SHapley Additive exPlanations) and LIME (Local Interpretable Model-Agnostic Explanations), the proposed system identifies behavioral patterns, skill gaps, and learning progress with transparency and fairness. The framework enhances educators’ ability to understand students’ coding approaches, decision-making patterns, and problemsolving efficiency. Experimental results show that integrating XAI methods enables better interpretability of predictive outcomes, thereby supporting personalized feedback and adaptive learning strategies.