The traditional physical education teaching mode of generalization gradually creates invisible limitations on the development of students as their physical fitness levels diversify. In this paper, association rule mining technology is selected to mine the association and hierarchical relationship between physical fitness data and sports performance data, which provides technical support for physical education teaching of students in higher vocational colleges and universities. The association rule mining algorithm framework is formed by deleting unnecessary data transaction items and improving the execution efficiency of Apriori algorithm. In addition, the idea of stratification is introduced to design a stratified training program and teaching planning with the backing of the data support of the physical fitness monitoring system. After the experimental class students were stratified and set up, all of them had the best improvement effect with the low level students (P<0.01), the middle level students (P<0.05), and the high level students (P<0.1). And both male and female groups showed different degrees of significant improvement in exercise level compared with the preexperiment (P<0.1). The combination of physical fitness monitoring and stratified training is a scientific and effective new sports teaching mode, which is guided by the real-time changes of students’ physical fitness data, and gives the most suitable sports training with full consideration of students’ individual differences, and assists in the improvement of students’ physical fitness level and sports performance.
