Digital Learning Behavior Analysis and Prediction Model Based on Deep Learning Algorithm

With the rapid development of digital education, learners' behavior data has become the key to the development of intelligent education. However, existing learning behavior analysis methods have problems such as low accuracy and poor generalization ability in predicting learning behavior trends and personalized recommendations. To this end, this paper models historical learning behavior data. The Transformer model can effectively identify learners' behavior patterns and predict key indicators such as learning progress, learning participation, and learning results. In addition, the model can also provide learners with personalized learning path recommendations, thereby improving learning outcomes and learners' enthusiasm. Experimental results show that the Transformer model performs well in multiple core evaluation indicators. Its hit rate @5 (Hit@5) reaches 71.40%, which is significantly higher than LSTM (63.50%), GRU (61.90%) and content-based recommendation methods (55.20%), indicating that Transformer can more accurately recommend content that learners are really interested in or are about to learn.