A Study on Differential Influence of Students’ Academic Performance on Stress Perception by Machine Learning

There are differences in the impact of students’ academic performance on perceived stress, and this difference may be related to students’ own and family attributes. Based on the principles of education, the impact function of academic performance on perceived stress was constructed as the dependent variable in this study. The data was collected to investigate academic performance perceived stress, self-attributes, and family attributes of students in grades 7–12 from a high school in Guangxi, China. A classification model for the impact function of academic performance on perceived stress was constructed using the decision tree method in machine learning. The experimental results showed that the F1-score of the decision tree model reached 85.6% with high accuracy. The experimental results validated the hypothesis and provided several attributes relevant to the impact of academic performance on perceived stress, as well as the portrait features of the most affected student group. The results of the decision tree model showed strong interpretability and provided a reference for teaching practice.