Influencing Factors of Physical Activity in Chinese University Students Based on Random Forest

Abstract Objective: This study investigates the key factors influencing physical activity levels among Chinese university students, employing a random forest algorithm to analyze the predictive power and relative importance of various variables. Methods: A cross-sectional study utilizing stratified cluster sampling was conducted, covering 17 provinces across China, and collecting 10,182 valid questionnaires. Physical activity levels were assessed using the International Physical Activity Questionnaire (IPAQ). A random forest algorithm was then used to analyze the importance of 39 variables in influencing physical activity. Results: Random Forest prediction showed that Exercise Adherence (Exercise Behavior), Exercise Adherence Level, Sex, and Exercise Adherence (Effort Investment) are the most significant factors affecting PA levels in university students. Mastery of Sports Skills, Exercise Motivation (Ability), Alcohol Consumption Level, Exercise Adherence (Emotional Experience), Exercise Motivation (Social), and Exercise Motivation (Fun) are other important influencing factors. The model achieved an accuracy of 0.704 and an AUC value of 0.760, indicating good predictive performance. Conclusion: Exercise Adherence, Sex, Mastery of Sports Skills, Alcohol Consumption Level, and Exercise Motivation may influence PA levels in university students. When conducting sports activities, attention should be paid to enhancing the “emotional value” and social attributes of university students participating in PA, focusing on the exercise intentions of female students, and emphasizing the mastery of more sports skills.