USING E-LEARNING FACTORS TO PREDICT STUDENT PERFORMANCE IN THE PRACTICE OF PRECISION EDUCATION

ABSTRACT

The emergence of Covid-19 has accelerated digital transformation, while simultaneously disrupting traditional education. The implementation of distance education practices produced a massive amount of data generated by the deployed learning management systems. Educational data mining tools using machine learning methods can produce thorough student-level insights into what has become known as precision education. This study aims to investigate a convenient approach to analyze a data set of 480 students in the Middle East using three supervised machine learning methods (artificial neural networks, decision trees, and Naive Bayes) to predict overall performance using SPSS. The findings indicate that the naive Bayes algorithm achieved the highest accuracy of 89.85%, while the artificial neural networks algorithm achieved the lowest variance, with a standard deviation of 2.37. Besides, there are more valuable insights beyond accuracy that other Machine learning models can provide in the SPSS environment, such as visual representation and normalized importance of independent variables. Moreover, in the context of missing student data, the data set was evaluated if e-learning parameters alone can predict student performance. The findings suggest that e-learning parameters alone can predict student performance with an average accuracy of 84.49%. This study contributes to limit grade inflation in the age of online learning due to educational malpractices.