Analysis the Influence of Online Media on College Students' Bedtime Procrastination Based on Random Forest Regression

Abstract

This study combined randomized forest regression with traditional statistical methods to examine the association between bedtime procrastination and online media use among college students. A total of 1,268 students in Guangdong Province completed questionnaires including the Bedtime Procrastination Scale, the Online Social Support Scale, and the Mobile Phone Addiction Index. Pearson's correlation, multivariate stepwise regression, and random forest regression were used to explore the variable relationships and influencing factors. As an integrated learning method, random forest regression has strong performance in modeling nonlinear, high-dimensional data and capturing complex interactions while reducing the risk of overfitting. The results showed a significant positive correlation between cell phone addiction and bedtime procrastination (r=0.365,∗p∗<0.01), as well as a significant negative correlation between online social support and cell phone addiction (r=0.278,∗p∗<0.01). Random forest regression performed best in modeling the effect of cell phone addiction on bedtime procrastination (MSE=0.62, R2=0.19), explaining 19.1% of the variance. Stepwise regression further identified major, academic year, self-rated health, academic stress, and cell phone addiction as key predictors. we found that bedtime procrastination is quite common, particularly among third-year students and those with poor health, high academic stress, or severe phone addiction. The use of random forest regression effectively uncovered behavioral patterns, providing valuable support for targeted interventions aimed at reducing media overuse and improving sleep health.