A Personalized Math Learning Experience with Clustering and Random Forest Algorithms

Educational methodologies are evolving alongside technological advancement by adopting digital education systems. To increase user engagement and ensure that content aligns with individual needs and learning processes, recommendation systems have been integrated into these platforms. The MathE platform was developed as part of the digital transformation in education. This platform provides personalized assessment and educational content across 22 higher education-level mathematic topics. The recommendation system within MathE customizes content and evaluations based on student needs and expertise by employing clustering for question difficulty, graph-based learning path optimization, and a Random Forest model for dynamic question selection. To evaluate the impact of the developed recommendation system, student assessment results were compared between random question generation and the use of the recommendation system. This study focuses on two commonly used subtopics in MathE: Partial Differentiation and Matrices and Determinants, with 20-22 students per subtopic. Results indicate improvement in student performance: a 6.94% increase for Partial Differentiation and a 9.37% increase for Matrices and Determinants. These findings highlight the recommendation system’s effectiveness in enhancing student performance, contributing to a more personalized and efficient learning experience.