E-learning System Development In Existence Of Machine Learning Tools With Prediction Tasks

This paper provides a new machine learning-based technique for evaluating eLearning system usability. To construct prediction models, three machine learning methods (support vector machines, neural networks, and decision trees) are integrated with multiple linear regression to reveal the underlying link between an eLearning system\'s overall usability and its predictor qualities. A sensitivity analysis is then used to assess the predictors\' significance. Using both sensitivity settings and usability scores, a statistic called the severity index is constructed. A Pareto-like technique is used to arrange the severity index values, and the most important usability aspects are chosen. The case study\'s findings show that implementing the proposed technique enhances eLearning system diagnostics by identifying the most significant usability features. The proposed method might provide crucial information to usability experts on which measures could be improved to improve system usability for a certain group of eLearning system end-users.