MOOC's Student Results Classification by Comparing PNN and other Classifiers with Features Selection

Abstract

An urgent necessity during year 2020, it became a must that all universities around the world to move from traditional classrooms, COVID-19 epidemic forced schools and universities to change their plans by e-learning strategy and/or hosting Massive Open Online Courses (MOOCs). Since dropouts and failure rates of MOOCs' students is a well noticed problem, this paper proposes a new methodology in classifying students' results throughout MOOCs modules. By using Open University Learning Analytics Dataset (OULAD) and applying modern machine learning techniques, it becomes more useful to monitor factors affecting student performance and achievement. The proposed methodology contributed a new model that uses various feature selection algorithms and various classification algorithms including Probabilistic Neural Network (PNN) and other classification algorithms. Results showed that using certain feature selection algorithms in combination with PNN resulted in enhancing trend exploration and accuracy.