Exploring the value of computer data analysis: a study on the transformation of public physical education teaching mode in universities

Technological, social, and cultural developments are continuously reshaping education, highlighting the need for improved athletic and physical education (PE) training methods. This study investigates the role of computer data analysis in enhancing public PE teaching models in colleges to improve student outcomes. A total of 180 college students participated, with data collected through surveys and movement-based video recordings. Data preprocessing included Z-score normalization and histogram equalization. Feature extraction was performed using Principal Component Analysis (PCA) and Histogram of Oriented Gradients (HOG) to identify numerical patterns and visual movement features for injury prevention. A Bi-directional Long Short-Term Memory (Bi-LSTM) network was employed for performance prediction and real-time feedback, while a Faster Region-based Convolutional Neural Network (FRCNN) enabled movement detection, object tracking, and duration analysis. The proposed system allows dynamic evaluation and personalized recommendations. Experimental results demonstrate improved engagement, exercise frequency, and performance, achieving 98% accuracy, 95% precision, 95.02% recall, 85% engagement, 90% satisfaction, and 70% motivation, with a 0.99 correlation between engagement and motivation. These findings confirm the effectiveness of AI-driven data analysis in modern public PE systems.