There is a close correlation between students'online learning behaviors and their grades,and in order to explore the implicit relationship between students'learning behaviors,the study adopts an improved association rule mining algorithm to analyze their correlation.The experimental results show that the algorithm designed in the study performs better in terms of precision and recall,and runs faster than the traditional association rule mining algorithm,with a time reduction of 26.49%.The method realizes the analysis and correlation of learning behaviors from multi-dimensions,which provides scientific support and guidance for teaching decisions and research.