Learning behaviour analysis of online course learners using EEG and facial expression data

Abstract

The technology advancement and pandemic situation moved the world from board teaching to online for various courses and millions of students are opting online mode in reality. Therefore, it is essential to analyse the quality of online education by tracking the learning behaviour of online course learners. In traditional method, the test scores and learners' feedback are used to assess the online learning behaviour. However this is not an efficient approach because the knowledge gained by the process of online learning depends not only on the test scores but also on the learners’ attention during courses. Therefore the test score, attention score and facial expressions of the learner are the three types of data required to effectively measure the performance, attentiveness and emotional state of the learner respectively. In order to obtain more insights about the learning behaviour, the proposed research focuses on collecting the data from Electroencephalography (EEG), facial expressions and test scores and analyzing using machine learning and deep learning models. The model uses techniques like Random Forest Algorithm to find the test score, Recurrent Neural Network (RNN) to predict time series attention scores of the learner for a video lecture and Convolutional Neural Network (CNN) to classify into different emotions for the facial expression images obtained from the recording of online courses. The predicted results are test scores, attention scores and overall emotional state of the learner during online class. From this research outcome, it is inferred that the test score is proportional to the attention score of the learner in most cases. Analyzing the attention variation during the different videos reveals that there are more fluctuations of attention in lecture video in comparison to the entertainment video reveals that the learners showed happier emotions for entertainment video than the lecture video.