A Machine Learning-Based Adaptive Feedback System To Enhance Programming Skills Using Computational Thinking

Abstract

In the modern education landscape, where programming is essential across diverse domains, students often face challenges in applying programming logic to real-world problems despite having theoretical knowledge. This paper introduces a machine learning-based adaptive feedback system that evaluates and improves students' Computational Thinking (CT) skills—essential for solving algorithmic problems. The CT components include decomposition, abstraction, pattern recognition, and algorithm design. By using CT-aligned programming assessments and clustering students using the K-Means algorithm, the system groups learners based on similar skill levels and provides personalized feedback. This feedback loop is designed to strengthen weak areas and reinforce cognitive development in programming. Results demonstrated that over 82% of students improved significantly in targeted CT areas. The approach promotes intelligent, scalable, and personalized education—redefining how we assess and support learners in coding education.