MOOC Dropout Prediction Based on Dynamic Embedding Representation Learning

Massive Open Online Courses (MOOCs) received great attentions in recent years. Most MOOCs have huge number of participants, which usually introduce another challenge—the extremely high dropout rate. Thus, people use a large amount of user-item interaction data collected from the MOOC platform to predict the dropout behaviors for further analysis. Dynamic embedding representation learning presents an attractive opportunity to model the dynamic evolution of users and items, where each user (item) can be embedded in a Euclidean space. This article introduces and analyzes the application of the joint dynamic user-item embedding algorithm in the MOOC dropout prediction. The empirical results indicated that the model has low dependence on data volume. Moreover, the model is robust to label-flipping attacks. Therefore, we believe that the model performances under different settings can be used to guide the real-world MOOC dropout prediction.