A study of stratified optimization of college physical education course effects under the perspective of cluster analysis

Due to the different physical fitness of college students, they have different acceptance abilities for college physical education courses, so a unified teaching method cannot be applied among many students. The article focuses on the student stratification model based on learning characteristics and takes the university physical education course as an example to design a university physical education teaching stratification method based on the improved K-means clustering algorithm with initialization denoising and center-of-mass optimization. Taking the physical education test data of the 2021~2023 sessions of a university as the research object, K-means clustering analysis was carried out on the 8-item physical fitness test data to realize the stratification of students’ learning situation. The experimental results show that the method of this paper clusters 14619 students into three layers, and the cluster centroid distances of Cluster 1~3 are 20.35, 16.97, and 18.39, respectively. The maximum standardized mutual information values are 6.820, 3.105, and 2.931, respectively, and the results of the stratification are reasonable and easy to interpret, and they can provide effective references for the development of personalized physical education for school students.