Student behaviour analysis and innovative curriculum planning based on apriori-HGNN model

This study proposes a learning situation analysis and curriculum planning model that integrates apriori algorithm and heterogeneous graph neural network to address the shortcomings of traditional methods in capturing long-term spatiotemporal dependencies, high-dimensional sparse features, and dynamic adaptability of campus data.The experiment shows that this method achieves an accuracy of 98.26% ± 0.15 in course recommendation, significantly better than comparative models such as collaborative filtering, matrix factorisation, factorisation machine, graph convolutional network, and reinforcement learning (p < 0.001).The rationality score of its course planning reached 9.62, the long-term learning effect improved by 26.95%, and the standardised cumulative benefit index was the highest (0.927 ± 0.006).The research results have achieved collaborative optimisation of behaviour pattern discovery and course semantic reasoning, achieving a good balance between accuracy and interpretability, and providing effective support for intelligent education decision-making.