Behavioral prediction and health benefit evaluation of leisure sports activities based on multi-physiological characteristics analysis

Leisure sports activities, as a healthy form of entertainment, have garnered increasing recognition. This paper introduces a data analysis model designed for behavior prediction and health benefit evaluation in leisure sports activities, utilizing multiple physiological features. The proposed model offers recommendations for leisure sports activities and provides health assessment results based on an array of physiological feature data. Constructed using a combination of Lasso (Least Absolute Shrinkage and Selection Operator) and GBDT (Gradient Boosting Decision Tree) regression models within the Stacking ensemble learning framework, the model leverages physiological feature data from the dataset for training. Experimental results reveal that the combined prediction model achieves a coefficient of determination of 0.9832, effectively mitigating the impact of pathological data on model fitting and demonstrating superior accuracy and stability compared to individual prediction models. Finally, this paper explores the future prospects of wearable devices for physiological feature data collection and the potential advancements in behavior prediction and health benefit evaluation methods based on such information.