MOOCs Dropout Prediction via Classmates Augmented Time-Flow Hybrid Network

Massive Open Online Courses (MOOCs) provide learners with a platform for free learning. However, MOOCs have been criticized for high dropout rates in recent years. For the purpose of predicting users’ potential dropout risk in advance, a novel framework named Classmates Augmented Time-Flow Hybrid Network (CA-TFHN) is proposed in this paper. TFHN, which takes advantage of LSTM and Self-Attention mechanism, is designed to generate user activity features by using user learning records. At the same time, an effective correlation calculation is defined based on user potential interests on courses with link prediction, bringing in relationships of classmates. Influences among classmates, modeled by a reconstructed user graph, are employed to augment the activity features of the user, resulting in an accurate prediction of dropout. Experiments on the XuetangX dataset demonstrate the effectiveness of CA-TFHN in predicting MOOCs dropout. The CA-TFHN codes are available from https://github.com/codeds27/CA-TFHN .