Grade Prediction Based On Neural Network Cascade Modeling

With the upgrading of big data governance in colleges and universities, the lack of grade prediction accuracy in colleges and universities is more prominent. In order to further explore the deep association between students’ multidimensional behavioral data and grades, and to realize the accurate prediction of students’ grades in colleges and universities, this paper constructs a cascade prediction model integrating the attention mechanism of multi-task graph neural network AMGNN and BP-G neural network. The multidimensional behavioral data of students in a university is used as the basic data, and its association with grades is mined, and the multidimensional comparison is verified by traditional machine learning algorithms, BP neural networks and constructed algorithms, and the results show that the newly constructed cascade model has a higher prediction accuracy in predicting the results. The use of this model can realize accurate grade prediction, provide data support for college teachers to improve teaching quality, and help college teaching management to develop to a higher level.