Predicting Practical Performance of the Student from C Code Submissions

Abstract

Understanding and predicting student outcomes in programming education requires innovative approaches that analyze code submissions as indicators of learning proficiency. This paper investigates the potential of programming feature analysis to evaluate student performance by examining C code submissions. Metrics such as lines of code, use of comments, variable types, condition checks, loops, syntax errors, and output formatting are explored to uncover patterns that reflect students' coding proficiency and conceptual understanding. The study presents a detailed review of feature extraction techniques, data preprocessing methods, and analytical frameworks, offering insights into the relationship between code quality and educational outcomes. By enabling data-driven insights, this research supports personalized learning strategies and enhances decision-making in programming education.