Research on the Learning Patterns and Error Pattern Recognition of College Students’ Sports Skills Based on Behavior Sequence Mining

Sports skill learning is a core component of college physical education and follows specific physiological and psychological laws. Timely recognition and correction of erroneous movements are essential for improving teaching quality, yet traditional teaching methods often show low efficiency and insufficient personalization. This study constructs a behavior-sequence-mining framework to explore learning patterns and identify error patterns in college students’ sports skill acquisition. Experiments are conducted on basketball shooting and football shooting skills among 300 college students. Motion capture, video recording, and expert annotation are used to collect multidimensional learning behavior data, including joint angles, movement trajectories, practice frequency, and error frequency. After preprocessing, the PrefixSpan algorithm is applied to mine frequent phased behavior sequences, and a support vector machine model is used to classify error patterns. The results show that the proposed method effectively mines stage-specific sequence patterns and achieves 91.8% accuracy in error pattern recognition, outperforming traditional manual recognition methods. The approach can accurately locate typical error types and provide data-driven support for personalized physical education guidance.