Prediction and Classification of Learning Styles using Machine Learning Approach

Abstract

E-platforms one of the most important solutions whose services must be provided for the education process to be more efficient and the fastest way to create an educational connection between professors and learners, in particularly in the difficult circumstances left by the Covid-19 pandemic. The e-learning or the new learning paradigm is a method that has imposed itself on universities for years, and it is a new form of education that is developing with the development of technologies. The e-learning is essentially an innovative type of self-service; the learner learns alone without the help of his teacher. In other words, the learner orients himself towards this service with his will and his choice, because he selects the service, which seems to him to meet his needs and his aspirations. Moreover, in learning each learner has different learning strategies, called learning style means the methods, techniques or systems designed to increase the efficiency and improve the level of learning, and develop students' scientific and cognitive abilities in various activities and tasks. So all learners must be brought to this awareness, it is necessary to offer educational activities of different types in order to respond to the different learning profiles of learners. Accordingly, this study presents the appropriate strategy that must be followed for the success of e-learning in E-platforms. By analyzing the responses of learners in our dataset and comparing the performance in predicting the learning styles of machine learning algorithms NN, RF, DT, BN, ANN and SVM. To determine the strongest algorithm and the most suitable for E-platforms, based on the most popular styles captured in the V ARK model (visual, auditory, read/write, and kinesthetic). The results of this study showed that the NN algorithm is the most accurate (98,7%) and efficient ones in predicting the learning style oflearners.