KMAN Publication Inc. (KMANPUB) Journal of Adolescent and Youth Psychological Studies (JAYPS) 2981-2526 7 Serial Number 46 2026 06 10 LightGBM-Based Prediction of Academic Burnout among High School Students from Perfectionism, Test Anxiety, Academic Self-Efficacy, and School Climate Indicators LightGBM-Based Prediction of Academic Burnout among High School Students from Perfectionism, Test Anxiety, Academic Self-Efficacy, and School Climate Indicators 1 11 10.61838/kman.jayps.5458 EN Journal Article 2026 02 12 <table> <tbody> <tr> <td> <p><strong>Objective:</strong> To develop and evaluate a machine learning model using LightGBM to predict academic burnout among high school students based on psychological and school environment factors.</p> <p><strong>Methods and Materials: </strong>A total of 1,248 high school students from Denmark participated in this cross-sectional study. Data were collected on perfectionism, test anxiety, academic self-efficacy, and perceptions of school climate using validated self-report instruments. Academic burnout was measured with the Maslach Burnout Inventory–Student Survey. Demographic data including age, gender, grade level, and school type were also obtained. Data were cleaned, standardized, and analyzed using Python and LightGBM algorithms. The dataset was divided into training (80%) and testing (20%) sets, and five-fold cross-validation was employed for model optimization. Model performance was assessed using accuracy, precision, recall, F1-score, ROC-AUC, and feature importance via SHAP values.</p> <p><strong>Findings: </strong>The LightGBM model accurately classified students into low- and high-risk burnout categories with 89.1% accuracy on the testing dataset. Test anxiety emerged as the strongest predictor, followed by academic self-efficacy, school climate, and perfectionism. Higher test anxiety and perfectionism were associated with increased burnout risk, whereas higher self-efficacy and more positive school climate perceptions were protective. The SHAP summary plot revealed the individual contributions of each predictor and indicated heterogeneity in their effects across students. Demographic variables contributed minimally to predictive performance.</p> <p><strong>Conclusion: </strong>This study demonstrates the effectiveness of a LightGBM-based model for predicting academic burnout using psychological and environmental factors. Findings emphasize the importance of addressing test anxiety, promoting academic self-efficacy, and fostering supportive school climates to reduce burnout risk among adolescents. Machine learning approaches can inform targeted interventions and early identification of at-risk students.</p> </td> </tr> </tbody> </table> https://journals.kmanpub.com/index.php/jayps/article/download/5458/10600