Using Deep Learning Techniques to Predict the level of E-learning Students by Analyzing Their Behavior

Abstract

Educational institutions in our current era are moving towards using artificial intelligence techniques, because of the important services they can provide in the field of education in general and in the field of e-learning in particular, as they help in evaluating the performance of the institution and determining the needs and requirements of students, which reflects positively on the performance of the educational institution and its various cadres. Therefore, educational facilities are competing to build systems with high accuracy to predict students’ needs in order to address them and avoid their academic failure. In this research, we propose ensemble classifier to predict students’ results based on their classroom behavior and demographic information, composed of several deep learning models capable of predicting accurately and reliably. The proposed system consists of a stacking classifier consisting of three sub classifiers, which are the extra decision tree classifiers and the logistic regression classifier. The study was able to reach a prediction rate of 87.5% for students’ level.

Keywords

• online learning
• deep learning
• ensemble classifier
• extra trees classifier
• logistic regression classifier
• stacking classifier.