Predicting the academic performance of students with GPcSAGE

Abstract

Educational data mining is a popular research area in data mining, and predicting student performance is one of the important research topics in educational data mining. In order to predict student performance in a timely and accurate manner, this paper proposes a Graph Pearson correlation Sample and AggreGatE (GPcSAGE) model based on graph neural networks. The sampling probability of neighboring nodes similar to the target node is optimized to weaken the influence of abnormal target node attributes on the prediction results and reduce the sampling variance. The algorithm efficiency and prediction accuracy are improved by reconfiguring the aggregation function to aggregate more important information. The experiments demonstrate the effectiveness of the method, which helps to predict students' learning trends and effects for precise teaching interventions to improve teaching quality.