Cluster Analysis of College Students' Physical Education Behavior Based on a Hybrid Algorithm of K-means and DBSCAN

<jats:p>Insight into students' athletic behavior patterns is critical to efficient resource deployment and enhanced engagement within higher education. This study develops a clustering method for university student sports behavior based on a hybrid algorithm of K-means and Density-Based Spatial Clustering of Applications with Noise (DBSCAN). The hybrid algorithm first uses K-means for data pre-segmentation to identify dense core regions, then uses DBSCAN to fine-tune cluster boundaries. Experimental data come from a university's smart sports platform, covering the exercise behavior records of 5,000 to 10,000 students over an academic year, including features such as exercise frequency, single-session duration, and activity preferences. The hybrid algorithm achieved a silhouette coefficient of 0.71 and successfully identified four typical groups with 89.3% accuracy. Compared with the single K-means and DBSCAN algorithms, this hybrid method improves both clustering accuracy and noise handling capabilities. The research results provide data support for the design of personalized sports curricula and dynamic venue scheduling.</jats:p>