With the advancement of society and the improvement of living standards, the significance of students’ physical health has become increasingly prominent. However, currently, the assessment and analysis of students’ physical health heavily depend on conventional statistical methods. Even with the application of data mining-related methodologies for analysis and evaluation, the exploitation and utilization of physical health big data remain relatively restricted. In this paper, an improved Canopy-K-means algorithm based on principal component analysis (PCA) is used to construct and analyze a portrait of students’ physical fitness and health. The method combines data dimensionality reduction techniques and cluster analysis techniques, and its combined performance is the best compared to other algorithms in the ablation experiments for both male and female student data groups. In this paper, the algorithm was used to process the grouping of physical fitness test data of male and female students, realize the construction and analysis of students’ physical fitness and health portrait, and give the exercise prescription for different categories of students. In this paper, the physical health test data of students of Yunnan Agricultural University in 2020–2022 were collected to carry out experiments, and the results found that there are differences in physical health status among students of different genders, grades, and majors in this university, and the physical health status of the students of Classes 2018 and 2019 is generally deficient; on different majors, the students of the Faculty of Agricultural Sciences are slightly superior to the Faculty of Science and Technology, and the students of the Faculty of Science and Technology are slightly superior to the students of the Faculty of Humanities and Social Sciences. Our study offers novel methods and ideas for the assessment and analysis of students’ physical health, holding significant implications for schools and related departments in formulating scientific and effective physical education policies and health promotion strategies.

随着社会的进步和生活水平的提高，学生身体健康的重要性日益凸显。然而，目前对学生身体健康状况的评估和分析很大程度上依赖于传统的统计方法。即使应用数据挖掘相关方法进行分析和评估，身体健康大数据的开发和利用仍然相对有限。本文采用基于主成分分析（PCA）的改进Canopy- K -means算法来构建和分析学生体能和健康状况的画像。该方法结合了数据降维技术和聚类分析技术，在男女学生数据组的消融实验中，与其他算法相比，其组合性能是最好的。本文利用该算法对男女学生体质测试数据进行分组处理，实现学生体质健康画像的构建与分析，并针对不同类别的学生给出运动处方。本文收集了云南农业大学2020—2022年学生体质健康测试数据进行实验，结果发现该校不同性别、年级、专业的学生体质健康状况存在差异。大学，2018、2019级学生身体健康状况普遍欠佳；不同专业上，农学院学生略优于科技学院，科技学院学生略优于人文社会科学学院学生。本研究为学生身体健康状况的评估和分析提供了新颖的方法和思路，对于学校及相关部门制定科学有效的体育教育政策和健康促进策略具有重要意义。