Random forest model combined with LASSO regression to analyze the influencing factors of aggressive behavior among high school students
10.16835/j.cnki.1000-9817.2025292
- VernacularTitle:随机森林模型联合LASSO回归分析高中生攻击性行为影响因素
- Author:
LI Yaxin, HU Qibi, WANG Xingyu, LIU Xin,CHANG Hongjuan
1
Author Information
1. School of Medicine, Wuhan University of Science and Technology, Wuhan 430000, Hubei Province, China
- Publication Type:Journal Article
- Keywords:
Models,statistical;
Regression analysis;
Aggression;
Students
- From:
Chinese Journal of School Health
2025;46(10):1437-1442
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To analyze the influencing factors of aggressive behaviors in high school students, rank the importance of relevant variables with visual presentation, so as to provide new references for developing intervention strategies.
Methods:In December 2024, a convenience sampling method was used to select 5 330 students from 5 high schools in Wuhan for an online questionnaire survey. The survey included the Buss-Perry Aggression Questionnaire (BPAQ), Interaction Anxiety Scale (IAS), Emotion Regulation Ability Scale, Inventory of Parent and Peer Attachment (IPPA), and a self designed general information questionnaire. Variables were screened using LASSO regression, followed by multivariate Logistic regression analysis. The Random Forest model was then employed to rank and visualize variable importance.
Results:Univariate analysis and LASSO regression results indicated that gender, emotion regulation capability, paternal attachment quality, maternal attachment quality, maternal parenting style, paternal parenting style, parenting consistency, monthly household income per capita, school climate, relationship with classmates, interaction anxiety, and academic ranking were all significant variables (all P <0.01). Multivariate Logistic regression analysis identified the following as influencing factors of aggressive behaviors in high school students: emotional management ability ( OR=0.38, 95%CI =0.34-0.43), gender ( OR=0.83, 95%CI =0.75-0.93), mother attachment quality ( OR=0.89, 95%CI = 0.74 -1.07), father attachment quality ( OR=0.76, 95%CI = 0.66- 0.87), interaction anxiousness ( OR=1.58, 95%CI =1.40- 1.78 ), consistency of parenting styles of parents ( OR=1.40, 95%CI =1.21-1.62), school atmosphere ( OR=1.50, 95%CI =1.14-1.98), academic performance ranking (upper middle: OR=0.76, 95%CI =0.61-0.93; middle: OR=0.70, 95%CI =0.57-0.86) were the influencing factors of aggressive behavior in high school students (all P <0.05). The results of Random Forest showed that emotional management ability was the most important factor affecting aggressive behavior in high school students.
Conclusion:Emotion regulation capability is identified as the core predictor of adolescent aggressive behaviors, which can significantly reduce the incidence of aggressive behaviors through enhanced emotion management.
