Learning Behavior Analysis and Dropout Rate Prediction Based on MOOCs Data

Abstract

With the continuous development of the MOOC, a large number of learners have joined the online classroom. Distance education has the advantage of being free from time and geographical restrictions. However, it still faces the dilemma of high dropout rate and the continuous loss of learners. By studying the MOOC log data, we model the various behaviors of students and hope to make more accurate predictions of dropout rates. The student s learning sequence information is essentially time-series data, and the time interval between events is often different, which leads to difficulties in prediction. Therefore, we propose a time-controlled Long Short-Term Memory neural network (E-LSTM) prediction model that incorporates time-control units, the unit has the ability to model early learning behaviors with different time intervals. Based on the original LSTM model, we design time-controlled gates to better capture long-and short-term information and simulate learning process information to improve forecast performance. The experimental results on the real MOOC dataset show that the accuracy of the proposed model is higher than that of multiple comparison models, which proves the effectiveness of the proposed method.