Abstract
Learning Analytics aims to discover the class of students' performance over time. This helps instructors make in-time interventions but, discovering the students' performance class in virtual learning environments consider a challenge due to distance constraints. Many studies, which applied to Massive Open Online Courses (MOOC) datasets, built predictive models but, these models were applied to specific courses and students and classify students into binary classes. Moreover, their results were obtained at the end of the course period thus delaying making in-time interventions. To bridge this gap, this study proposes a day-wise multi-class model to predict students' performance using Artificial Neural Network and Long Short-Term Memory, named ANN-LSTM. To check the validity of this model, two baseline models, the Recurrent Neural Network (RNN) and Gated Recurrent Unit (GRU), were conducted and compared with ANN-LSTM in this context. Additionally, the results of ANN-LSTM were compared with the state-of-the-art models in terms of accuracy. The results show that the ANN-LSTM model obtained the best results among baseline models. The accuracy obtained by ANN-LSTM was about 70% at the end of the third month of the course and outperforms RNN and GRU models which obtained 53% and 57%, respectively. Also, the ANN-LSTM model obtained the best accuracy results with enhancement rates of about 6-14% when compared with state-of-the-art models. This highlights the ability of LSTM as a time series model to make early predictions for student performance in MOOC taking benefit of its architecture and ability to keep latent dependencies.