Study on Learning Effect Prediction Based on MOOC Big Data

Learning behavior can reflect the learner's learning status and learning effect. In the era of rapid development of the Massive Open Online Courses (MOOC) learning platform, many researchers predict the learning effect based on the learner's learning behavior or participation, but few studies based on the learning behavior sequence data. This paper proposes an online education learning effect prediction model-MLEP-WDL (MOOC Learning Effect Prediction based on Wide & Deep Learning), which based on the learning behavior sequence data of the MOOC platform. This model considers the influence of sequence data on the prediction results, and adds a transformer structure to replace the commonly used CNN & RNN structure. The experimental results show that the F1 value in the prediction performance of the MLEP-WDL model reaches 85.74%. Compared with the benchmark model (CNN + LSTM and CNN + GRU), the improvement of the performance is more than 2%, which effectively confirms the feasibility and superiority of the model in learning effect prediction.