A prediction model for higher education learning behaviors based on deep learning
Abstract
With the rapid advancement of deep learning technology, research on intelligent models centered on learning behavior prediction has become a key direction in higher education informatization. Addressing the challenges of complex and highly temporal student behavior patterns in university online learning environments, this study proposes a deep learning model tailored for predicting learning behaviors in higher education. The model employs multi-layer neural network technology to perform deep data mining learning through data preprocessing, feature extraction, model construction, and optimization training, thereby achieving precise identification and dynamic prediction of learning behaviors. This paper utilizes data from a university online learning platform, encompassing information such as assignment submissions, resource access, and interactive participation. Experimental results demonstrate that this learning behavior prediction model achieves higher prediction accuracy (93%), recall rate (91%), and F1 score (92%) compared to LSTM, while reducing training time by nearly 30.7%. This validates the model's effectiveness in feature fusion and parameter optimization. As an intelligent learning analytics solution, deep learning-based learning behavior prediction models can serve as an effective technical approach for university learning support systems and personalized teaching decision-making.