A Study of College Students’ Academic Anxiety Modeling Based on Interpretable Machine Learning Algorithms

Purpose: Academic anxiety is a common phenomenon in the college student population, which has an important impact on students’ psychological health and academic performance. Therefore, by exploring the effects of college students’ professional commitment and achievement goal orientation variables on academic anxiety, it helps to understand students’ motivation and goal setting, so as to provide targeted academic guidance and assistance and help students better cope with academic anxiety. Methods: In this paper, the Professional Commitment Scale for College Students, the Achievement Goal Orientation Scale, and the Academic Anxiety Scale were used to conduct a questionnaire survey on 1534 college students. Based on the survey data, the Random Forest algorithm was used to construct a student anxiety model, and the SHAP method was used to analyze the feature interpretability of the indicators affecting the evaluation of college students’ academic anxiety. Results: The average prediction accuracy of RF-SHAP model for student anxiety reaches 97%, with the relatively highest contribution of mastery of avoidance goal orientation. Conclusion: In this paper, the machine learning algorithm model is applied to the analysis of academic anxiety, and high accuracy prediction effect is realized. By developing targeted intervention strategies through the perspectives of professional commitment and goal orientation, college students can reinterpret and reduce their sense of academic anxiety and develop the ability and confidence to cope with academic challenges.