Prediction of E-Learning Efficiency by Deep Learning in EKhool Online Portal Networks

Abstract

Nowadays, e-learning plays a very significant role in a global education system. E-learning is an effective way, which offers convenient and more benefits to users. In addition, it provides flexible learning and provides various contents based on the user’s needs. However, identification of student’s performance is a difficult one, because of virtual and dynamic environments. In this paper, Deep Belief Network (DBN) is used for predicting e-learning efficiency in e-khool online portal network. At first, key performance indicators, like the student’s profile, student’s behaviour data, student’s performance, and teacher’s involvement are determined for the prediction. Here, the student’s performance, like name, age, family income, sex, and location, whereas the student’s behaviour data, which includes the lecture hours, time spent on the discussion forum, attended exams, and a number of course subscribed are considered. Likewise, student’s performance, such as average exam score and average time spent in the exam room, whereas teacher’s involvement, such as the number of lectures given by teachers and the number of exams conducted by teachers are considered. Thus, the above-mentioned attributes are determined for each student. Then, the E-khool learning management system is introduced for extracting the performance indicators for further processing. Once the performance indicators are extracted, the deep learning model is applied for predicting the student performance using E-khool model. Here, the performance metrics, like Mean Absolute Error (MAE), accuracy, and Mean Square Error (MSE) are evaluated for analyzing the effectiveness of the developed method. Moreover, the DBN method obtains a maximum accuracy of 0.92% and also achieved a minimum MSE of 0.41% and a minimum MAE of 0.14%.

References